{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.995008319467554,
  "eval_steps": 500,
  "global_step": 1350,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022185246810870773,
      "grad_norm": 0.0481039322912693,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.6911,
      "step": 1
    },
    {
      "epoch": 0.004437049362174155,
      "grad_norm": 0.06187029927968979,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.598,
      "step": 2
    },
    {
      "epoch": 0.0066555740432612314,
      "grad_norm": 0.07853534072637558,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.6839,
      "step": 3
    },
    {
      "epoch": 0.00887409872434831,
      "grad_norm": 0.047561150044202805,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.5322,
      "step": 4
    },
    {
      "epoch": 0.011092623405435386,
      "grad_norm": 0.04162918031215668,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.5908,
      "step": 5
    },
    {
      "epoch": 0.013311148086522463,
      "grad_norm": 0.049715809524059296,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.5487,
      "step": 6
    },
    {
      "epoch": 0.01552967276760954,
      "grad_norm": 0.0693173035979271,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.5222,
      "step": 7
    },
    {
      "epoch": 0.01774819744869662,
      "grad_norm": 0.057486798614263535,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.7035,
      "step": 8
    },
    {
      "epoch": 0.019966722129783693,
      "grad_norm": 0.11299107223749161,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.5513,
      "step": 9
    },
    {
      "epoch": 0.022185246810870772,
      "grad_norm": 0.033717624843120575,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.7484,
      "step": 10
    },
    {
      "epoch": 0.024403771491957847,
      "grad_norm": 0.1958114355802536,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.9169,
      "step": 11
    },
    {
      "epoch": 0.026622296173044926,
      "grad_norm": 0.04956788569688797,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.5452,
      "step": 12
    },
    {
      "epoch": 0.028840820854132,
      "grad_norm": 0.05320743843913078,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.6818,
      "step": 13
    },
    {
      "epoch": 0.03105934553521908,
      "grad_norm": 0.037530381232500076,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.8535,
      "step": 14
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 0.07759865373373032,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.6085,
      "step": 15
    },
    {
      "epoch": 0.03549639489739324,
      "grad_norm": 0.05210991948843002,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.6864,
      "step": 16
    },
    {
      "epoch": 0.03771491957848031,
      "grad_norm": 0.06853218376636505,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.6783,
      "step": 17
    },
    {
      "epoch": 0.03993344425956739,
      "grad_norm": 0.0703229308128357,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.5744,
      "step": 18
    },
    {
      "epoch": 0.04215196894065446,
      "grad_norm": 0.04483000934123993,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.5662,
      "step": 19
    },
    {
      "epoch": 0.044370493621741544,
      "grad_norm": 0.05514465272426605,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.6061,
      "step": 20
    },
    {
      "epoch": 0.04658901830282862,
      "grad_norm": 0.05506841465830803,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.6804,
      "step": 21
    },
    {
      "epoch": 0.048807542983915694,
      "grad_norm": 0.0424894355237484,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.6613,
      "step": 22
    },
    {
      "epoch": 0.051026067665002776,
      "grad_norm": 0.06383293122053146,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.7025,
      "step": 23
    },
    {
      "epoch": 0.05324459234608985,
      "grad_norm": 0.032453689724206924,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.5338,
      "step": 24
    },
    {
      "epoch": 0.05546311702717693,
      "grad_norm": 0.07119202613830566,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.7401,
      "step": 25
    },
    {
      "epoch": 0.057681641708264,
      "grad_norm": 0.07367593795061111,
      "learning_rate": 5.777777777777777e-05,
      "loss": 0.6614,
      "step": 26
    },
    {
      "epoch": 0.059900166389351084,
      "grad_norm": 0.04507562890648842,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.7912,
      "step": 27
    },
    {
      "epoch": 0.06211869107043816,
      "grad_norm": 0.06193961948156357,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.5286,
      "step": 28
    },
    {
      "epoch": 0.06433721575152523,
      "grad_norm": 0.06222601607441902,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.709,
      "step": 29
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 0.06487862020730972,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.6017,
      "step": 30
    },
    {
      "epoch": 0.06877426511369938,
      "grad_norm": 0.04030263423919678,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.6183,
      "step": 31
    },
    {
      "epoch": 0.07099278979478647,
      "grad_norm": 0.05400936305522919,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.7484,
      "step": 32
    },
    {
      "epoch": 0.07321131447587355,
      "grad_norm": 0.055917274206876755,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.6761,
      "step": 33
    },
    {
      "epoch": 0.07542983915696062,
      "grad_norm": 0.06991104781627655,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.7045,
      "step": 34
    },
    {
      "epoch": 0.0776483638380477,
      "grad_norm": 0.06677992641925812,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.6048,
      "step": 35
    },
    {
      "epoch": 0.07986688851913477,
      "grad_norm": 0.04458651691675186,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.6422,
      "step": 36
    },
    {
      "epoch": 0.08208541320022185,
      "grad_norm": 0.04437301680445671,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.7224,
      "step": 37
    },
    {
      "epoch": 0.08430393788130892,
      "grad_norm": 0.06594880670309067,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.7334,
      "step": 38
    },
    {
      "epoch": 0.08652246256239601,
      "grad_norm": 0.16524019837379456,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.7359,
      "step": 39
    },
    {
      "epoch": 0.08874098724348309,
      "grad_norm": 0.04549632966518402,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.6808,
      "step": 40
    },
    {
      "epoch": 0.09095951192457016,
      "grad_norm": 0.04918269068002701,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.6228,
      "step": 41
    },
    {
      "epoch": 0.09317803660565724,
      "grad_norm": 0.06044666841626167,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.6415,
      "step": 42
    },
    {
      "epoch": 0.09539656128674431,
      "grad_norm": 0.08786917477846146,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.7305,
      "step": 43
    },
    {
      "epoch": 0.09761508596783139,
      "grad_norm": 0.04982609674334526,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.6376,
      "step": 44
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 0.06757058948278427,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.7119,
      "step": 45
    },
    {
      "epoch": 0.10205213533000555,
      "grad_norm": 0.14583416283130646,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.6556,
      "step": 46
    },
    {
      "epoch": 0.10427066001109263,
      "grad_norm": 0.07092802226543427,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.7898,
      "step": 47
    },
    {
      "epoch": 0.1064891846921797,
      "grad_norm": 0.05988693982362747,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.6374,
      "step": 48
    },
    {
      "epoch": 0.10870770937326678,
      "grad_norm": 0.05437200143933296,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.7634,
      "step": 49
    },
    {
      "epoch": 0.11092623405435385,
      "grad_norm": 0.07328523695468903,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.6855,
      "step": 50
    },
    {
      "epoch": 0.11314475873544093,
      "grad_norm": 0.062422335147857666,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.7096,
      "step": 51
    },
    {
      "epoch": 0.115363283416528,
      "grad_norm": 0.038714416325092316,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.7244,
      "step": 52
    },
    {
      "epoch": 0.11758180809761509,
      "grad_norm": 0.05603143200278282,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.7841,
      "step": 53
    },
    {
      "epoch": 0.11980033277870217,
      "grad_norm": 0.04840129613876343,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.789,
      "step": 54
    },
    {
      "epoch": 0.12201885745978924,
      "grad_norm": 0.039549730718135834,
      "learning_rate": 0.00012222222222222221,
      "loss": 0.5151,
      "step": 55
    },
    {
      "epoch": 0.12423738214087632,
      "grad_norm": 0.04703560471534729,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.5992,
      "step": 56
    },
    {
      "epoch": 0.1264559068219634,
      "grad_norm": 0.051590416580438614,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.5769,
      "step": 57
    },
    {
      "epoch": 0.12867443150305047,
      "grad_norm": 0.059639811515808105,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.6643,
      "step": 58
    },
    {
      "epoch": 0.13089295618413754,
      "grad_norm": 0.05333787202835083,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.5987,
      "step": 59
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 0.04241842031478882,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.4581,
      "step": 60
    },
    {
      "epoch": 0.1353300055463117,
      "grad_norm": 0.050560712814331055,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.6981,
      "step": 61
    },
    {
      "epoch": 0.13754853022739877,
      "grad_norm": 0.0471394918859005,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.6242,
      "step": 62
    },
    {
      "epoch": 0.13976705490848584,
      "grad_norm": 0.0387752503156662,
      "learning_rate": 0.00014,
      "loss": 0.6382,
      "step": 63
    },
    {
      "epoch": 0.14198557958957295,
      "grad_norm": 0.0472797155380249,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.5812,
      "step": 64
    },
    {
      "epoch": 0.14420410427066002,
      "grad_norm": 0.06413045525550842,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.7331,
      "step": 65
    },
    {
      "epoch": 0.1464226289517471,
      "grad_norm": 0.043547432869672775,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.5887,
      "step": 66
    },
    {
      "epoch": 0.14864115363283417,
      "grad_norm": 0.057281244546175,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.6717,
      "step": 67
    },
    {
      "epoch": 0.15085967831392125,
      "grad_norm": 0.03370499610900879,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.4639,
      "step": 68
    },
    {
      "epoch": 0.15307820299500832,
      "grad_norm": 0.040247198194265366,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.6492,
      "step": 69
    },
    {
      "epoch": 0.1552967276760954,
      "grad_norm": 0.06381718814373016,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.641,
      "step": 70
    },
    {
      "epoch": 0.15751525235718247,
      "grad_norm": 0.037510525435209274,
      "learning_rate": 0.00015777777777777776,
      "loss": 0.5827,
      "step": 71
    },
    {
      "epoch": 0.15973377703826955,
      "grad_norm": 0.06351568549871445,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.6497,
      "step": 72
    },
    {
      "epoch": 0.16195230171935662,
      "grad_norm": 0.05052623525261879,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.613,
      "step": 73
    },
    {
      "epoch": 0.1641708264004437,
      "grad_norm": 0.056964609771966934,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.6059,
      "step": 74
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 0.060783203691244125,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.6421,
      "step": 75
    },
    {
      "epoch": 0.16860787576261785,
      "grad_norm": 0.0393964983522892,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.837,
      "step": 76
    },
    {
      "epoch": 0.17082640044370492,
      "grad_norm": 0.05392806604504585,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.7027,
      "step": 77
    },
    {
      "epoch": 0.17304492512479203,
      "grad_norm": 0.03672610595822334,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.5516,
      "step": 78
    },
    {
      "epoch": 0.1752634498058791,
      "grad_norm": 0.0598294772207737,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.708,
      "step": 79
    },
    {
      "epoch": 0.17748197448696618,
      "grad_norm": 0.04885537922382355,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.6276,
      "step": 80
    },
    {
      "epoch": 0.17970049916805325,
      "grad_norm": 0.05532560124993324,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.6663,
      "step": 81
    },
    {
      "epoch": 0.18191902384914033,
      "grad_norm": 0.03075404092669487,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.5697,
      "step": 82
    },
    {
      "epoch": 0.1841375485302274,
      "grad_norm": 0.08292128890752792,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.6982,
      "step": 83
    },
    {
      "epoch": 0.18635607321131448,
      "grad_norm": 0.044025447219610214,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.5346,
      "step": 84
    },
    {
      "epoch": 0.18857459789240155,
      "grad_norm": 0.14476321637630463,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.7677,
      "step": 85
    },
    {
      "epoch": 0.19079312257348863,
      "grad_norm": 0.04437650740146637,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.5642,
      "step": 86
    },
    {
      "epoch": 0.1930116472545757,
      "grad_norm": 0.05506033077836037,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.6699,
      "step": 87
    },
    {
      "epoch": 0.19523017193566278,
      "grad_norm": 0.04561001807451248,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.6549,
      "step": 88
    },
    {
      "epoch": 0.19744869661674985,
      "grad_norm": 0.06246626004576683,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.6544,
      "step": 89
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 0.0492669939994812,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.6802,
      "step": 90
    },
    {
      "epoch": 0.201885745978924,
      "grad_norm": 0.05632823705673218,
      "learning_rate": 0.0002022222222222222,
      "loss": 0.8799,
      "step": 91
    },
    {
      "epoch": 0.2041042706600111,
      "grad_norm": 0.04394519701600075,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.6407,
      "step": 92
    },
    {
      "epoch": 0.20632279534109818,
      "grad_norm": 0.05078943446278572,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.7509,
      "step": 93
    },
    {
      "epoch": 0.20854132002218526,
      "grad_norm": 0.09347491711378098,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.6737,
      "step": 94
    },
    {
      "epoch": 0.21075984470327233,
      "grad_norm": 0.053966034203767776,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.699,
      "step": 95
    },
    {
      "epoch": 0.2129783693843594,
      "grad_norm": 0.047602977603673935,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.8016,
      "step": 96
    },
    {
      "epoch": 0.21519689406544648,
      "grad_norm": 0.04507390037178993,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.5755,
      "step": 97
    },
    {
      "epoch": 0.21741541874653356,
      "grad_norm": 0.03397609665989876,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.5914,
      "step": 98
    },
    {
      "epoch": 0.21963394342762063,
      "grad_norm": 0.047897424548864365,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.7492,
      "step": 99
    },
    {
      "epoch": 0.2218524681087077,
      "grad_norm": 0.049556851387023926,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.7904,
      "step": 100
    },
    {
      "epoch": 0.22407099278979478,
      "grad_norm": 0.05308236926794052,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.5727,
      "step": 101
    },
    {
      "epoch": 0.22628951747088186,
      "grad_norm": 0.036327335983514786,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.6443,
      "step": 102
    },
    {
      "epoch": 0.22850804215196893,
      "grad_norm": 0.043715376406908035,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.6542,
      "step": 103
    },
    {
      "epoch": 0.230726566833056,
      "grad_norm": 0.05818153917789459,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.6368,
      "step": 104
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 0.04389097914099693,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.6358,
      "step": 105
    },
    {
      "epoch": 0.23516361619523019,
      "grad_norm": 0.05562211573123932,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.5691,
      "step": 106
    },
    {
      "epoch": 0.23738214087631726,
      "grad_norm": 0.04758453741669655,
      "learning_rate": 0.00023777777777777775,
      "loss": 0.6218,
      "step": 107
    },
    {
      "epoch": 0.23960066555740434,
      "grad_norm": 0.037191227078437805,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.6352,
      "step": 108
    },
    {
      "epoch": 0.2418191902384914,
      "grad_norm": 0.03637460619211197,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.7153,
      "step": 109
    },
    {
      "epoch": 0.24403771491957849,
      "grad_norm": 0.04701869934797287,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.694,
      "step": 110
    },
    {
      "epoch": 0.24625623960066556,
      "grad_norm": 0.044285621494054794,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.4896,
      "step": 111
    },
    {
      "epoch": 0.24847476428175264,
      "grad_norm": 0.05454069375991821,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.5855,
      "step": 112
    },
    {
      "epoch": 0.2506932889628397,
      "grad_norm": 0.0656687542796135,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.663,
      "step": 113
    },
    {
      "epoch": 0.2529118136439268,
      "grad_norm": 0.0577431358397007,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.7784,
      "step": 114
    },
    {
      "epoch": 0.25513033832501386,
      "grad_norm": 0.03564617037773132,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.6208,
      "step": 115
    },
    {
      "epoch": 0.25734886300610094,
      "grad_norm": 0.08875054121017456,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.7636,
      "step": 116
    },
    {
      "epoch": 0.259567387687188,
      "grad_norm": 0.06940072029829025,
      "learning_rate": 0.00026,
      "loss": 0.6997,
      "step": 117
    },
    {
      "epoch": 0.2617859123682751,
      "grad_norm": 0.047868210822343826,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.6118,
      "step": 118
    },
    {
      "epoch": 0.26400443704936216,
      "grad_norm": 0.055806249380111694,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.7476,
      "step": 119
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 0.12321016192436218,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.9201,
      "step": 120
    },
    {
      "epoch": 0.2684414864115363,
      "grad_norm": 0.06718350201845169,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.6381,
      "step": 121
    },
    {
      "epoch": 0.2706600110926234,
      "grad_norm": 0.048017892986536026,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.7364,
      "step": 122
    },
    {
      "epoch": 0.27287853577371046,
      "grad_norm": 0.043548889458179474,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.5342,
      "step": 123
    },
    {
      "epoch": 0.27509706045479754,
      "grad_norm": 0.04628445580601692,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.673,
      "step": 124
    },
    {
      "epoch": 0.2773155851358846,
      "grad_norm": 0.0453525148332119,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.6766,
      "step": 125
    },
    {
      "epoch": 0.2795341098169717,
      "grad_norm": 0.04236726835370064,
      "learning_rate": 0.00028,
      "loss": 0.51,
      "step": 126
    },
    {
      "epoch": 0.2817526344980588,
      "grad_norm": 0.05125441402196884,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.5769,
      "step": 127
    },
    {
      "epoch": 0.2839711591791459,
      "grad_norm": 0.032166119664907455,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.5848,
      "step": 128
    },
    {
      "epoch": 0.28618968386023297,
      "grad_norm": 0.04116509482264519,
      "learning_rate": 0.0002866666666666667,
      "loss": 0.7047,
      "step": 129
    },
    {
      "epoch": 0.28840820854132004,
      "grad_norm": 0.05057681351900101,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.645,
      "step": 130
    },
    {
      "epoch": 0.2906267332224071,
      "grad_norm": 0.063532255589962,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.7026,
      "step": 131
    },
    {
      "epoch": 0.2928452579034942,
      "grad_norm": 0.04839206859469414,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.5052,
      "step": 132
    },
    {
      "epoch": 0.29506378258458127,
      "grad_norm": 0.047344207763671875,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.7141,
      "step": 133
    },
    {
      "epoch": 0.29728230726566834,
      "grad_norm": 0.05061426758766174,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.5683,
      "step": 134
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 0.07713436335325241,
      "learning_rate": 0.0003,
      "loss": 0.5455,
      "step": 135
    },
    {
      "epoch": 0.3017193566278425,
      "grad_norm": 0.10872895270586014,
      "learning_rate": 0.00029975308641975307,
      "loss": 0.502,
      "step": 136
    },
    {
      "epoch": 0.30393788130892957,
      "grad_norm": 0.06562027335166931,
      "learning_rate": 0.00029950617283950616,
      "loss": 0.7946,
      "step": 137
    },
    {
      "epoch": 0.30615640599001664,
      "grad_norm": 0.04317990317940712,
      "learning_rate": 0.00029925925925925926,
      "loss": 0.6397,
      "step": 138
    },
    {
      "epoch": 0.3083749306711037,
      "grad_norm": 0.05150894075632095,
      "learning_rate": 0.0002990123456790123,
      "loss": 0.6059,
      "step": 139
    },
    {
      "epoch": 0.3105934553521908,
      "grad_norm": 0.04636860638856888,
      "learning_rate": 0.0002987654320987654,
      "loss": 0.6534,
      "step": 140
    },
    {
      "epoch": 0.31281198003327787,
      "grad_norm": 0.05661846697330475,
      "learning_rate": 0.0002985185185185185,
      "loss": 0.5765,
      "step": 141
    },
    {
      "epoch": 0.31503050471436495,
      "grad_norm": 0.05222322791814804,
      "learning_rate": 0.0002982716049382716,
      "loss": 0.6182,
      "step": 142
    },
    {
      "epoch": 0.317249029395452,
      "grad_norm": 0.0453859344124794,
      "learning_rate": 0.0002980246913580247,
      "loss": 0.6984,
      "step": 143
    },
    {
      "epoch": 0.3194675540765391,
      "grad_norm": 0.05608554556965828,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.5349,
      "step": 144
    },
    {
      "epoch": 0.32168607875762617,
      "grad_norm": 0.04674987122416496,
      "learning_rate": 0.00029753086419753087,
      "loss": 0.6141,
      "step": 145
    },
    {
      "epoch": 0.32390460343871325,
      "grad_norm": 0.043753620237112045,
      "learning_rate": 0.0002972839506172839,
      "loss": 0.6106,
      "step": 146
    },
    {
      "epoch": 0.3261231281198003,
      "grad_norm": 0.03406202793121338,
      "learning_rate": 0.000297037037037037,
      "loss": 0.4648,
      "step": 147
    },
    {
      "epoch": 0.3283416528008874,
      "grad_norm": 0.041685979813337326,
      "learning_rate": 0.0002967901234567901,
      "loss": 0.5512,
      "step": 148
    },
    {
      "epoch": 0.33056017748197447,
      "grad_norm": 0.07592085748910904,
      "learning_rate": 0.0002965432098765432,
      "loss": 0.6502,
      "step": 149
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.04438149929046631,
      "learning_rate": 0.00029629629629629624,
      "loss": 0.6226,
      "step": 150
    },
    {
      "epoch": 0.3349972268441486,
      "grad_norm": 0.034320928156375885,
      "learning_rate": 0.00029604938271604933,
      "loss": 0.5825,
      "step": 151
    },
    {
      "epoch": 0.3372157515252357,
      "grad_norm": 0.06777451932430267,
      "learning_rate": 0.0002958024691358025,
      "loss": 0.6615,
      "step": 152
    },
    {
      "epoch": 0.33943427620632277,
      "grad_norm": 0.04912026226520538,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.6545,
      "step": 153
    },
    {
      "epoch": 0.34165280088740985,
      "grad_norm": 0.05765862762928009,
      "learning_rate": 0.0002953086419753086,
      "loss": 0.6814,
      "step": 154
    },
    {
      "epoch": 0.343871325568497,
      "grad_norm": 0.05012679845094681,
      "learning_rate": 0.0002950617283950617,
      "loss": 0.6847,
      "step": 155
    },
    {
      "epoch": 0.34608985024958405,
      "grad_norm": 0.045678481459617615,
      "learning_rate": 0.0002948148148148148,
      "loss": 0.5718,
      "step": 156
    },
    {
      "epoch": 0.3483083749306711,
      "grad_norm": 0.08242905139923096,
      "learning_rate": 0.00029456790123456785,
      "loss": 0.6571,
      "step": 157
    },
    {
      "epoch": 0.3505268996117582,
      "grad_norm": 0.05195464566349983,
      "learning_rate": 0.00029432098765432095,
      "loss": 0.6545,
      "step": 158
    },
    {
      "epoch": 0.3527454242928453,
      "grad_norm": 0.04192949831485748,
      "learning_rate": 0.00029407407407407404,
      "loss": 0.5181,
      "step": 159
    },
    {
      "epoch": 0.35496394897393235,
      "grad_norm": 0.05133754387497902,
      "learning_rate": 0.00029382716049382714,
      "loss": 0.5178,
      "step": 160
    },
    {
      "epoch": 0.35718247365501943,
      "grad_norm": 0.041282299906015396,
      "learning_rate": 0.00029358024691358023,
      "loss": 0.5911,
      "step": 161
    },
    {
      "epoch": 0.3594009983361065,
      "grad_norm": 0.03932717815041542,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.6796,
      "step": 162
    },
    {
      "epoch": 0.3616195230171936,
      "grad_norm": 0.041161179542541504,
      "learning_rate": 0.0002930864197530864,
      "loss": 0.5989,
      "step": 163
    },
    {
      "epoch": 0.36383804769828065,
      "grad_norm": 0.03443289175629616,
      "learning_rate": 0.00029283950617283946,
      "loss": 0.5357,
      "step": 164
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.051769789308309555,
      "learning_rate": 0.00029259259259259256,
      "loss": 0.5428,
      "step": 165
    },
    {
      "epoch": 0.3682750970604548,
      "grad_norm": 0.05101555585861206,
      "learning_rate": 0.00029234567901234565,
      "loss": 0.6771,
      "step": 166
    },
    {
      "epoch": 0.3704936217415419,
      "grad_norm": 0.04635120928287506,
      "learning_rate": 0.00029209876543209875,
      "loss": 0.6126,
      "step": 167
    },
    {
      "epoch": 0.37271214642262895,
      "grad_norm": 0.03293093293905258,
      "learning_rate": 0.00029185185185185184,
      "loss": 0.5636,
      "step": 168
    },
    {
      "epoch": 0.37493067110371603,
      "grad_norm": 0.043570976704359055,
      "learning_rate": 0.0002916049382716049,
      "loss": 0.4885,
      "step": 169
    },
    {
      "epoch": 0.3771491957848031,
      "grad_norm": 0.045075275003910065,
      "learning_rate": 0.000291358024691358,
      "loss": 0.6547,
      "step": 170
    },
    {
      "epoch": 0.3793677204658902,
      "grad_norm": 0.04765894636511803,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.7044,
      "step": 171
    },
    {
      "epoch": 0.38158624514697725,
      "grad_norm": 0.048752810806035995,
      "learning_rate": 0.00029086419753086417,
      "loss": 0.5667,
      "step": 172
    },
    {
      "epoch": 0.38380476982806433,
      "grad_norm": 0.05341934785246849,
      "learning_rate": 0.00029061728395061726,
      "loss": 0.6758,
      "step": 173
    },
    {
      "epoch": 0.3860232945091514,
      "grad_norm": 0.05386456847190857,
      "learning_rate": 0.00029037037037037036,
      "loss": 0.697,
      "step": 174
    },
    {
      "epoch": 0.3882418191902385,
      "grad_norm": 0.11125757545232773,
      "learning_rate": 0.00029012345679012345,
      "loss": 0.5721,
      "step": 175
    },
    {
      "epoch": 0.39046034387132555,
      "grad_norm": 0.04296626150608063,
      "learning_rate": 0.0002898765432098765,
      "loss": 0.6059,
      "step": 176
    },
    {
      "epoch": 0.39267886855241263,
      "grad_norm": 0.05087501183152199,
      "learning_rate": 0.0002896296296296296,
      "loss": 0.6592,
      "step": 177
    },
    {
      "epoch": 0.3948973932334997,
      "grad_norm": 0.04276210442185402,
      "learning_rate": 0.0002893827160493827,
      "loss": 0.5842,
      "step": 178
    },
    {
      "epoch": 0.3971159179145868,
      "grad_norm": 0.03969917818903923,
      "learning_rate": 0.0002891358024691358,
      "loss": 0.6871,
      "step": 179
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 0.04886876046657562,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.628,
      "step": 180
    },
    {
      "epoch": 0.40155296727676093,
      "grad_norm": 0.05216556787490845,
      "learning_rate": 0.00028864197530864197,
      "loss": 0.6508,
      "step": 181
    },
    {
      "epoch": 0.403771491957848,
      "grad_norm": 0.04216371476650238,
      "learning_rate": 0.00028839506172839507,
      "loss": 0.5541,
      "step": 182
    },
    {
      "epoch": 0.4059900166389351,
      "grad_norm": 0.05433110520243645,
      "learning_rate": 0.0002881481481481481,
      "loss": 0.5764,
      "step": 183
    },
    {
      "epoch": 0.4082085413200222,
      "grad_norm": 0.04063545539975166,
      "learning_rate": 0.0002879012345679012,
      "loss": 0.6284,
      "step": 184
    },
    {
      "epoch": 0.4104270660011093,
      "grad_norm": 0.03528410941362381,
      "learning_rate": 0.0002876543209876543,
      "loss": 0.5323,
      "step": 185
    },
    {
      "epoch": 0.41264559068219636,
      "grad_norm": 0.05878869444131851,
      "learning_rate": 0.0002874074074074074,
      "loss": 0.6905,
      "step": 186
    },
    {
      "epoch": 0.41486411536328344,
      "grad_norm": 0.07532719522714615,
      "learning_rate": 0.00028716049382716043,
      "loss": 0.6452,
      "step": 187
    },
    {
      "epoch": 0.4170826400443705,
      "grad_norm": 0.039721522480249405,
      "learning_rate": 0.00028691358024691353,
      "loss": 0.5934,
      "step": 188
    },
    {
      "epoch": 0.4193011647254576,
      "grad_norm": 0.07444796711206436,
      "learning_rate": 0.0002866666666666667,
      "loss": 0.8041,
      "step": 189
    },
    {
      "epoch": 0.42151968940654466,
      "grad_norm": 0.04292351379990578,
      "learning_rate": 0.0002864197530864197,
      "loss": 0.5774,
      "step": 190
    },
    {
      "epoch": 0.42373821408763174,
      "grad_norm": 0.04220479726791382,
      "learning_rate": 0.0002861728395061728,
      "loss": 0.5208,
      "step": 191
    },
    {
      "epoch": 0.4259567387687188,
      "grad_norm": 0.04417847841978073,
      "learning_rate": 0.0002859259259259259,
      "loss": 0.653,
      "step": 192
    },
    {
      "epoch": 0.4281752634498059,
      "grad_norm": 0.03365249186754227,
      "learning_rate": 0.000285679012345679,
      "loss": 0.6183,
      "step": 193
    },
    {
      "epoch": 0.43039378813089296,
      "grad_norm": 0.04270458221435547,
      "learning_rate": 0.00028543209876543204,
      "loss": 0.5018,
      "step": 194
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 0.0517653226852417,
      "learning_rate": 0.00028518518518518514,
      "loss": 0.5318,
      "step": 195
    },
    {
      "epoch": 0.4348308374930671,
      "grad_norm": 0.06575360894203186,
      "learning_rate": 0.00028493827160493824,
      "loss": 0.8209,
      "step": 196
    },
    {
      "epoch": 0.4370493621741542,
      "grad_norm": 0.04742785543203354,
      "learning_rate": 0.00028469135802469133,
      "loss": 0.6927,
      "step": 197
    },
    {
      "epoch": 0.43926788685524126,
      "grad_norm": 0.047831423580646515,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.6095,
      "step": 198
    },
    {
      "epoch": 0.44148641153632834,
      "grad_norm": 0.04418404772877693,
      "learning_rate": 0.0002841975308641975,
      "loss": 0.6899,
      "step": 199
    },
    {
      "epoch": 0.4437049362174154,
      "grad_norm": 0.03292892128229141,
      "learning_rate": 0.0002839506172839506,
      "loss": 0.6359,
      "step": 200
    },
    {
      "epoch": 0.4459234608985025,
      "grad_norm": 0.04176653176546097,
      "learning_rate": 0.00028370370370370366,
      "loss": 0.6183,
      "step": 201
    },
    {
      "epoch": 0.44814198557958956,
      "grad_norm": 0.03740749880671501,
      "learning_rate": 0.00028345679012345675,
      "loss": 0.4966,
      "step": 202
    },
    {
      "epoch": 0.45036051026067664,
      "grad_norm": 0.037485163658857346,
      "learning_rate": 0.00028320987654320985,
      "loss": 0.6119,
      "step": 203
    },
    {
      "epoch": 0.4525790349417637,
      "grad_norm": 0.06673204898834229,
      "learning_rate": 0.00028296296296296294,
      "loss": 0.8415,
      "step": 204
    },
    {
      "epoch": 0.4547975596228508,
      "grad_norm": 0.08209707587957382,
      "learning_rate": 0.00028271604938271604,
      "loss": 0.7467,
      "step": 205
    },
    {
      "epoch": 0.45701608430393786,
      "grad_norm": 0.03921116888523102,
      "learning_rate": 0.0002824691358024691,
      "loss": 0.6131,
      "step": 206
    },
    {
      "epoch": 0.45923460898502494,
      "grad_norm": 0.04010651260614395,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.7199,
      "step": 207
    },
    {
      "epoch": 0.461453133666112,
      "grad_norm": 0.040419355034828186,
      "learning_rate": 0.00028197530864197527,
      "loss": 0.5758,
      "step": 208
    },
    {
      "epoch": 0.4636716583471991,
      "grad_norm": 0.07303260266780853,
      "learning_rate": 0.00028172839506172836,
      "loss": 0.4942,
      "step": 209
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 0.04907373711466789,
      "learning_rate": 0.00028148148148148146,
      "loss": 0.6565,
      "step": 210
    },
    {
      "epoch": 0.46810870770937324,
      "grad_norm": 0.039416342973709106,
      "learning_rate": 0.00028123456790123455,
      "loss": 0.5346,
      "step": 211
    },
    {
      "epoch": 0.47032723239046037,
      "grad_norm": 0.044889237731695175,
      "learning_rate": 0.00028098765432098765,
      "loss": 0.6355,
      "step": 212
    },
    {
      "epoch": 0.47254575707154745,
      "grad_norm": 0.043690748512744904,
      "learning_rate": 0.0002807407407407407,
      "loss": 0.6371,
      "step": 213
    },
    {
      "epoch": 0.4747642817526345,
      "grad_norm": 0.05129116401076317,
      "learning_rate": 0.0002804938271604938,
      "loss": 0.6085,
      "step": 214
    },
    {
      "epoch": 0.4769828064337216,
      "grad_norm": 0.043414484709501266,
      "learning_rate": 0.0002802469135802469,
      "loss": 0.6068,
      "step": 215
    },
    {
      "epoch": 0.47920133111480867,
      "grad_norm": 0.06400758773088455,
      "learning_rate": 0.00028,
      "loss": 0.6436,
      "step": 216
    },
    {
      "epoch": 0.48141985579589575,
      "grad_norm": 0.047464821487665176,
      "learning_rate": 0.00027975308641975307,
      "loss": 0.5819,
      "step": 217
    },
    {
      "epoch": 0.4836383804769828,
      "grad_norm": 0.04285236820578575,
      "learning_rate": 0.00027950617283950617,
      "loss": 0.5867,
      "step": 218
    },
    {
      "epoch": 0.4858569051580699,
      "grad_norm": 0.04382825642824173,
      "learning_rate": 0.00027925925925925926,
      "loss": 0.5752,
      "step": 219
    },
    {
      "epoch": 0.48807542983915697,
      "grad_norm": 0.04115068539977074,
      "learning_rate": 0.0002790123456790123,
      "loss": 0.5709,
      "step": 220
    },
    {
      "epoch": 0.49029395452024405,
      "grad_norm": 0.056331828236579895,
      "learning_rate": 0.0002787654320987654,
      "loss": 0.6906,
      "step": 221
    },
    {
      "epoch": 0.4925124792013311,
      "grad_norm": 0.04520461708307266,
      "learning_rate": 0.0002785185185185185,
      "loss": 0.6378,
      "step": 222
    },
    {
      "epoch": 0.4947310038824182,
      "grad_norm": 0.03697853907942772,
      "learning_rate": 0.0002782716049382716,
      "loss": 0.7174,
      "step": 223
    },
    {
      "epoch": 0.49694952856350527,
      "grad_norm": 0.04596490412950516,
      "learning_rate": 0.00027802469135802463,
      "loss": 0.6931,
      "step": 224
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.06441549956798553,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.588,
      "step": 225
    },
    {
      "epoch": 0.5013865779256794,
      "grad_norm": 0.06302216649055481,
      "learning_rate": 0.00027753086419753087,
      "loss": 0.7367,
      "step": 226
    },
    {
      "epoch": 0.5036051026067665,
      "grad_norm": 0.04973161593079567,
      "learning_rate": 0.0002772839506172839,
      "loss": 0.8054,
      "step": 227
    },
    {
      "epoch": 0.5058236272878536,
      "grad_norm": 0.03232715651392937,
      "learning_rate": 0.000277037037037037,
      "loss": 0.5852,
      "step": 228
    },
    {
      "epoch": 0.5080421519689406,
      "grad_norm": 0.09985991567373276,
      "learning_rate": 0.0002767901234567901,
      "loss": 0.682,
      "step": 229
    },
    {
      "epoch": 0.5102606766500277,
      "grad_norm": 0.03853890672326088,
      "learning_rate": 0.0002765432098765432,
      "loss": 0.6301,
      "step": 230
    },
    {
      "epoch": 0.5124792013311148,
      "grad_norm": 0.062123093754053116,
      "learning_rate": 0.00027629629629629624,
      "loss": 0.4914,
      "step": 231
    },
    {
      "epoch": 0.5146977260122019,
      "grad_norm": 0.04679717496037483,
      "learning_rate": 0.00027604938271604934,
      "loss": 0.6445,
      "step": 232
    },
    {
      "epoch": 0.516916250693289,
      "grad_norm": 0.04000224173069,
      "learning_rate": 0.00027580246913580243,
      "loss": 0.6648,
      "step": 233
    },
    {
      "epoch": 0.519134775374376,
      "grad_norm": 0.047567155212163925,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.5578,
      "step": 234
    },
    {
      "epoch": 0.5213533000554631,
      "grad_norm": 0.054090313613414764,
      "learning_rate": 0.0002753086419753086,
      "loss": 0.5032,
      "step": 235
    },
    {
      "epoch": 0.5235718247365502,
      "grad_norm": 0.03870861604809761,
      "learning_rate": 0.0002750617283950617,
      "loss": 0.6307,
      "step": 236
    },
    {
      "epoch": 0.5257903494176372,
      "grad_norm": 0.037884388118982315,
      "learning_rate": 0.0002748148148148148,
      "loss": 0.6105,
      "step": 237
    },
    {
      "epoch": 0.5280088740987243,
      "grad_norm": 0.0458051860332489,
      "learning_rate": 0.00027456790123456785,
      "loss": 0.6914,
      "step": 238
    },
    {
      "epoch": 0.5302273987798114,
      "grad_norm": 0.04943552985787392,
      "learning_rate": 0.00027432098765432095,
      "loss": 0.5467,
      "step": 239
    },
    {
      "epoch": 0.5324459234608985,
      "grad_norm": 0.044316865503787994,
      "learning_rate": 0.00027407407407407404,
      "loss": 0.539,
      "step": 240
    },
    {
      "epoch": 0.5346644481419855,
      "grad_norm": 0.069974385201931,
      "learning_rate": 0.00027382716049382714,
      "loss": 0.632,
      "step": 241
    },
    {
      "epoch": 0.5368829728230726,
      "grad_norm": 0.041703302413225174,
      "learning_rate": 0.00027358024691358023,
      "loss": 0.5216,
      "step": 242
    },
    {
      "epoch": 0.5391014975041597,
      "grad_norm": 0.035853270441293716,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.5156,
      "step": 243
    },
    {
      "epoch": 0.5413200221852468,
      "grad_norm": 0.036384738981723785,
      "learning_rate": 0.0002730864197530864,
      "loss": 0.5631,
      "step": 244
    },
    {
      "epoch": 0.5435385468663338,
      "grad_norm": 0.054006434977054596,
      "learning_rate": 0.00027283950617283946,
      "loss": 0.6376,
      "step": 245
    },
    {
      "epoch": 0.5457570715474209,
      "grad_norm": 0.05898372083902359,
      "learning_rate": 0.00027259259259259256,
      "loss": 0.4783,
      "step": 246
    },
    {
      "epoch": 0.547975596228508,
      "grad_norm": 0.03320194035768509,
      "learning_rate": 0.00027234567901234565,
      "loss": 0.6261,
      "step": 247
    },
    {
      "epoch": 0.5501941209095951,
      "grad_norm": 0.05233430117368698,
      "learning_rate": 0.00027209876543209875,
      "loss": 0.5061,
      "step": 248
    },
    {
      "epoch": 0.5524126455906821,
      "grad_norm": 0.03165731206536293,
      "learning_rate": 0.00027185185185185184,
      "loss": 0.5391,
      "step": 249
    },
    {
      "epoch": 0.5546311702717692,
      "grad_norm": 0.047696344554424286,
      "learning_rate": 0.0002716049382716049,
      "loss": 0.5415,
      "step": 250
    },
    {
      "epoch": 0.5568496949528563,
      "grad_norm": 0.04902045056223869,
      "learning_rate": 0.00027135802469135803,
      "loss": 0.6707,
      "step": 251
    },
    {
      "epoch": 0.5590682196339434,
      "grad_norm": 0.04169073700904846,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.5509,
      "step": 252
    },
    {
      "epoch": 0.5612867443150305,
      "grad_norm": 0.03650841861963272,
      "learning_rate": 0.00027086419753086417,
      "loss": 0.5868,
      "step": 253
    },
    {
      "epoch": 0.5635052689961176,
      "grad_norm": 0.04365259036421776,
      "learning_rate": 0.00027061728395061727,
      "loss": 0.7102,
      "step": 254
    },
    {
      "epoch": 0.5657237936772047,
      "grad_norm": 0.034259580075740814,
      "learning_rate": 0.00027037037037037036,
      "loss": 0.5471,
      "step": 255
    },
    {
      "epoch": 0.5679423183582918,
      "grad_norm": 0.04212673753499985,
      "learning_rate": 0.00027012345679012346,
      "loss": 0.5372,
      "step": 256
    },
    {
      "epoch": 0.5701608430393789,
      "grad_norm": 0.042296767234802246,
      "learning_rate": 0.0002698765432098765,
      "loss": 0.5912,
      "step": 257
    },
    {
      "epoch": 0.5723793677204659,
      "grad_norm": 0.08112405240535736,
      "learning_rate": 0.0002696296296296296,
      "loss": 0.7017,
      "step": 258
    },
    {
      "epoch": 0.574597892401553,
      "grad_norm": 0.04262432083487511,
      "learning_rate": 0.0002693827160493827,
      "loss": 0.5942,
      "step": 259
    },
    {
      "epoch": 0.5768164170826401,
      "grad_norm": 0.03243022784590721,
      "learning_rate": 0.0002691358024691358,
      "loss": 0.6149,
      "step": 260
    },
    {
      "epoch": 0.5790349417637272,
      "grad_norm": 0.0785943940281868,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.7227,
      "step": 261
    },
    {
      "epoch": 0.5812534664448142,
      "grad_norm": 0.04140807315707207,
      "learning_rate": 0.00026864197530864197,
      "loss": 0.4449,
      "step": 262
    },
    {
      "epoch": 0.5834719911259013,
      "grad_norm": 0.04396454617381096,
      "learning_rate": 0.000268395061728395,
      "loss": 0.5939,
      "step": 263
    },
    {
      "epoch": 0.5856905158069884,
      "grad_norm": 0.05384218692779541,
      "learning_rate": 0.0002681481481481481,
      "loss": 0.5261,
      "step": 264
    },
    {
      "epoch": 0.5879090404880755,
      "grad_norm": 0.04736911877989769,
      "learning_rate": 0.0002679012345679012,
      "loss": 0.5611,
      "step": 265
    },
    {
      "epoch": 0.5901275651691625,
      "grad_norm": 0.044167596846818924,
      "learning_rate": 0.0002676543209876543,
      "loss": 0.6889,
      "step": 266
    },
    {
      "epoch": 0.5923460898502496,
      "grad_norm": 0.040374599397182465,
      "learning_rate": 0.0002674074074074074,
      "loss": 0.6244,
      "step": 267
    },
    {
      "epoch": 0.5945646145313367,
      "grad_norm": 0.05354900658130646,
      "learning_rate": 0.00026716049382716043,
      "loss": 0.6827,
      "step": 268
    },
    {
      "epoch": 0.5967831392124238,
      "grad_norm": 0.04152179881930351,
      "learning_rate": 0.0002669135802469136,
      "loss": 0.7265,
      "step": 269
    },
    {
      "epoch": 0.5990016638935108,
      "grad_norm": 0.043077029287815094,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.6527,
      "step": 270
    },
    {
      "epoch": 0.6012201885745979,
      "grad_norm": 0.05530163273215294,
      "learning_rate": 0.0002664197530864197,
      "loss": 0.6639,
      "step": 271
    },
    {
      "epoch": 0.603438713255685,
      "grad_norm": 0.0378279946744442,
      "learning_rate": 0.0002661728395061728,
      "loss": 0.4504,
      "step": 272
    },
    {
      "epoch": 0.6056572379367721,
      "grad_norm": 0.06108143553137779,
      "learning_rate": 0.0002659259259259259,
      "loss": 0.639,
      "step": 273
    },
    {
      "epoch": 0.6078757626178591,
      "grad_norm": 0.037403956055641174,
      "learning_rate": 0.000265679012345679,
      "loss": 0.7244,
      "step": 274
    },
    {
      "epoch": 0.6100942872989462,
      "grad_norm": 0.030271945521235466,
      "learning_rate": 0.00026543209876543205,
      "loss": 0.4551,
      "step": 275
    },
    {
      "epoch": 0.6123128119800333,
      "grad_norm": 0.05491235479712486,
      "learning_rate": 0.00026518518518518514,
      "loss": 0.6286,
      "step": 276
    },
    {
      "epoch": 0.6145313366611204,
      "grad_norm": 0.0368659682571888,
      "learning_rate": 0.00026493827160493824,
      "loss": 0.6078,
      "step": 277
    },
    {
      "epoch": 0.6167498613422074,
      "grad_norm": 0.05600963905453682,
      "learning_rate": 0.00026469135802469133,
      "loss": 0.5475,
      "step": 278
    },
    {
      "epoch": 0.6189683860232945,
      "grad_norm": 0.047419048845767975,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.5289,
      "step": 279
    },
    {
      "epoch": 0.6211869107043816,
      "grad_norm": 0.05446114018559456,
      "learning_rate": 0.0002641975308641975,
      "loss": 0.5099,
      "step": 280
    },
    {
      "epoch": 0.6234054353854687,
      "grad_norm": 0.04911911115050316,
      "learning_rate": 0.0002639506172839506,
      "loss": 0.548,
      "step": 281
    },
    {
      "epoch": 0.6256239600665557,
      "grad_norm": 0.04629703611135483,
      "learning_rate": 0.00026370370370370366,
      "loss": 0.5363,
      "step": 282
    },
    {
      "epoch": 0.6278424847476428,
      "grad_norm": 0.027827316895127296,
      "learning_rate": 0.00026345679012345675,
      "loss": 0.4272,
      "step": 283
    },
    {
      "epoch": 0.6300610094287299,
      "grad_norm": 0.035215821117162704,
      "learning_rate": 0.00026320987654320985,
      "loss": 0.5225,
      "step": 284
    },
    {
      "epoch": 0.632279534109817,
      "grad_norm": 0.04645289480686188,
      "learning_rate": 0.00026296296296296294,
      "loss": 0.5968,
      "step": 285
    },
    {
      "epoch": 0.634498058790904,
      "grad_norm": 0.04772736877202988,
      "learning_rate": 0.00026271604938271604,
      "loss": 0.5817,
      "step": 286
    },
    {
      "epoch": 0.6367165834719911,
      "grad_norm": 0.04018162935972214,
      "learning_rate": 0.00026246913580246913,
      "loss": 0.5851,
      "step": 287
    },
    {
      "epoch": 0.6389351081530782,
      "grad_norm": 0.0787653997540474,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.5774,
      "step": 288
    },
    {
      "epoch": 0.6411536328341653,
      "grad_norm": 0.03931837156414986,
      "learning_rate": 0.00026197530864197527,
      "loss": 0.6602,
      "step": 289
    },
    {
      "epoch": 0.6433721575152523,
      "grad_norm": 0.08393155038356781,
      "learning_rate": 0.00026172839506172837,
      "loss": 0.7284,
      "step": 290
    },
    {
      "epoch": 0.6455906821963394,
      "grad_norm": 0.046530622988939285,
      "learning_rate": 0.00026148148148148146,
      "loss": 0.6797,
      "step": 291
    },
    {
      "epoch": 0.6478092068774265,
      "grad_norm": 0.03741439804434776,
      "learning_rate": 0.00026123456790123456,
      "loss": 0.641,
      "step": 292
    },
    {
      "epoch": 0.6500277315585136,
      "grad_norm": 0.04188067838549614,
      "learning_rate": 0.0002609876543209876,
      "loss": 0.5976,
      "step": 293
    },
    {
      "epoch": 0.6522462562396006,
      "grad_norm": 0.041353873908519745,
      "learning_rate": 0.0002607407407407407,
      "loss": 0.5536,
      "step": 294
    },
    {
      "epoch": 0.6544647809206877,
      "grad_norm": 0.04282671958208084,
      "learning_rate": 0.00026049382716049384,
      "loss": 0.467,
      "step": 295
    },
    {
      "epoch": 0.6566833056017748,
      "grad_norm": 0.04725685715675354,
      "learning_rate": 0.0002602469135802469,
      "loss": 0.6892,
      "step": 296
    },
    {
      "epoch": 0.6589018302828619,
      "grad_norm": 0.036562807857990265,
      "learning_rate": 0.00026,
      "loss": 0.5074,
      "step": 297
    },
    {
      "epoch": 0.6611203549639489,
      "grad_norm": 0.037142783403396606,
      "learning_rate": 0.00025975308641975307,
      "loss": 0.5349,
      "step": 298
    },
    {
      "epoch": 0.663338879645036,
      "grad_norm": 0.04593242332339287,
      "learning_rate": 0.00025950617283950617,
      "loss": 0.7602,
      "step": 299
    },
    {
      "epoch": 0.6655574043261231,
      "grad_norm": 0.04612836614251137,
      "learning_rate": 0.0002592592592592592,
      "loss": 0.697,
      "step": 300
    },
    {
      "epoch": 0.6677759290072102,
      "grad_norm": 0.05162101611495018,
      "learning_rate": 0.0002590123456790123,
      "loss": 0.5401,
      "step": 301
    },
    {
      "epoch": 0.6699944536882972,
      "grad_norm": 0.04960690811276436,
      "learning_rate": 0.0002587654320987654,
      "loss": 0.6217,
      "step": 302
    },
    {
      "epoch": 0.6722129783693843,
      "grad_norm": 0.03715214878320694,
      "learning_rate": 0.0002585185185185185,
      "loss": 0.5315,
      "step": 303
    },
    {
      "epoch": 0.6744315030504714,
      "grad_norm": 0.048218853771686554,
      "learning_rate": 0.0002582716049382716,
      "loss": 0.4798,
      "step": 304
    },
    {
      "epoch": 0.6766500277315585,
      "grad_norm": 0.03557295352220535,
      "learning_rate": 0.0002580246913580247,
      "loss": 0.6079,
      "step": 305
    },
    {
      "epoch": 0.6788685524126455,
      "grad_norm": 0.04065033793449402,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.5144,
      "step": 306
    },
    {
      "epoch": 0.6810870770937326,
      "grad_norm": 0.05322769284248352,
      "learning_rate": 0.0002575308641975308,
      "loss": 0.6472,
      "step": 307
    },
    {
      "epoch": 0.6833056017748197,
      "grad_norm": 0.04282677173614502,
      "learning_rate": 0.0002572839506172839,
      "loss": 0.6657,
      "step": 308
    },
    {
      "epoch": 0.6855241264559068,
      "grad_norm": 0.04395861178636551,
      "learning_rate": 0.000257037037037037,
      "loss": 0.5549,
      "step": 309
    },
    {
      "epoch": 0.687742651136994,
      "grad_norm": 0.044052522629499435,
      "learning_rate": 0.0002567901234567901,
      "loss": 0.4863,
      "step": 310
    },
    {
      "epoch": 0.689961175818081,
      "grad_norm": 0.048362504690885544,
      "learning_rate": 0.0002565432098765432,
      "loss": 0.6993,
      "step": 311
    },
    {
      "epoch": 0.6921797004991681,
      "grad_norm": 0.039040155708789825,
      "learning_rate": 0.00025629629629629624,
      "loss": 0.6018,
      "step": 312
    },
    {
      "epoch": 0.6943982251802552,
      "grad_norm": 0.03936184570193291,
      "learning_rate": 0.0002560493827160494,
      "loss": 0.5591,
      "step": 313
    },
    {
      "epoch": 0.6966167498613423,
      "grad_norm": 0.06213848665356636,
      "learning_rate": 0.00025580246913580243,
      "loss": 0.6351,
      "step": 314
    },
    {
      "epoch": 0.6988352745424293,
      "grad_norm": 0.033867113292217255,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.5664,
      "step": 315
    },
    {
      "epoch": 0.7010537992235164,
      "grad_norm": 0.047558367252349854,
      "learning_rate": 0.0002553086419753086,
      "loss": 0.5252,
      "step": 316
    },
    {
      "epoch": 0.7032723239046035,
      "grad_norm": 0.03367559611797333,
      "learning_rate": 0.0002550617283950617,
      "loss": 0.4376,
      "step": 317
    },
    {
      "epoch": 0.7054908485856906,
      "grad_norm": 0.043320104479789734,
      "learning_rate": 0.0002548148148148148,
      "loss": 0.4215,
      "step": 318
    },
    {
      "epoch": 0.7077093732667776,
      "grad_norm": 0.03888561949133873,
      "learning_rate": 0.00025456790123456785,
      "loss": 0.6228,
      "step": 319
    },
    {
      "epoch": 0.7099278979478647,
      "grad_norm": 0.04222287982702255,
      "learning_rate": 0.00025432098765432095,
      "loss": 0.689,
      "step": 320
    },
    {
      "epoch": 0.7121464226289518,
      "grad_norm": 0.037576355040073395,
      "learning_rate": 0.00025407407407407404,
      "loss": 0.5144,
      "step": 321
    },
    {
      "epoch": 0.7143649473100389,
      "grad_norm": 0.05184217169880867,
      "learning_rate": 0.00025382716049382714,
      "loss": 0.6356,
      "step": 322
    },
    {
      "epoch": 0.7165834719911259,
      "grad_norm": 0.04148878529667854,
      "learning_rate": 0.00025358024691358023,
      "loss": 0.566,
      "step": 323
    },
    {
      "epoch": 0.718801996672213,
      "grad_norm": 0.08165695518255234,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.5959,
      "step": 324
    },
    {
      "epoch": 0.7210205213533001,
      "grad_norm": 0.04304583743214607,
      "learning_rate": 0.0002530864197530864,
      "loss": 0.4873,
      "step": 325
    },
    {
      "epoch": 0.7232390460343872,
      "grad_norm": 0.04006177559494972,
      "learning_rate": 0.00025283950617283947,
      "loss": 0.5968,
      "step": 326
    },
    {
      "epoch": 0.7254575707154742,
      "grad_norm": 0.0422949343919754,
      "learning_rate": 0.00025259259259259256,
      "loss": 0.6464,
      "step": 327
    },
    {
      "epoch": 0.7276760953965613,
      "grad_norm": 0.05775041878223419,
      "learning_rate": 0.00025234567901234566,
      "loss": 0.6384,
      "step": 328
    },
    {
      "epoch": 0.7298946200776484,
      "grad_norm": 0.04126613214612007,
      "learning_rate": 0.00025209876543209875,
      "loss": 0.509,
      "step": 329
    },
    {
      "epoch": 0.7321131447587355,
      "grad_norm": 0.043026942759752274,
      "learning_rate": 0.0002518518518518518,
      "loss": 0.5689,
      "step": 330
    },
    {
      "epoch": 0.7343316694398225,
      "grad_norm": 0.04095824062824249,
      "learning_rate": 0.00025160493827160494,
      "loss": 0.7086,
      "step": 331
    },
    {
      "epoch": 0.7365501941209096,
      "grad_norm": 0.042397063225507736,
      "learning_rate": 0.00025135802469135804,
      "loss": 0.5937,
      "step": 332
    },
    {
      "epoch": 0.7387687188019967,
      "grad_norm": 0.03201206400990486,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.4797,
      "step": 333
    },
    {
      "epoch": 0.7409872434830838,
      "grad_norm": 0.05407959967851639,
      "learning_rate": 0.00025086419753086417,
      "loss": 0.6283,
      "step": 334
    },
    {
      "epoch": 0.7432057681641708,
      "grad_norm": 0.05080745369195938,
      "learning_rate": 0.00025061728395061727,
      "loss": 0.622,
      "step": 335
    },
    {
      "epoch": 0.7454242928452579,
      "grad_norm": 0.04192512482404709,
      "learning_rate": 0.00025037037037037036,
      "loss": 0.6309,
      "step": 336
    },
    {
      "epoch": 0.747642817526345,
      "grad_norm": 0.04762016981840134,
      "learning_rate": 0.0002501234567901234,
      "loss": 0.6628,
      "step": 337
    },
    {
      "epoch": 0.7498613422074321,
      "grad_norm": 0.0408601313829422,
      "learning_rate": 0.0002498765432098765,
      "loss": 0.6365,
      "step": 338
    },
    {
      "epoch": 0.7520798668885191,
      "grad_norm": 0.05005708336830139,
      "learning_rate": 0.00024962962962962965,
      "loss": 0.6645,
      "step": 339
    },
    {
      "epoch": 0.7542983915696062,
      "grad_norm": 0.046575646847486496,
      "learning_rate": 0.0002493827160493827,
      "loss": 0.6921,
      "step": 340
    },
    {
      "epoch": 0.7565169162506933,
      "grad_norm": 0.07361260056495667,
      "learning_rate": 0.0002491358024691358,
      "loss": 0.5959,
      "step": 341
    },
    {
      "epoch": 0.7587354409317804,
      "grad_norm": 0.05646577849984169,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.4548,
      "step": 342
    },
    {
      "epoch": 0.7609539656128674,
      "grad_norm": 0.04482068121433258,
      "learning_rate": 0.000248641975308642,
      "loss": 0.4563,
      "step": 343
    },
    {
      "epoch": 0.7631724902939545,
      "grad_norm": 0.051413487643003464,
      "learning_rate": 0.000248395061728395,
      "loss": 0.6416,
      "step": 344
    },
    {
      "epoch": 0.7653910149750416,
      "grad_norm": 0.06049666553735733,
      "learning_rate": 0.0002481481481481481,
      "loss": 0.6403,
      "step": 345
    },
    {
      "epoch": 0.7676095396561287,
      "grad_norm": 0.03501538187265396,
      "learning_rate": 0.0002479012345679012,
      "loss": 0.5971,
      "step": 346
    },
    {
      "epoch": 0.7698280643372157,
      "grad_norm": 0.04038376733660698,
      "learning_rate": 0.0002476543209876543,
      "loss": 0.531,
      "step": 347
    },
    {
      "epoch": 0.7720465890183028,
      "grad_norm": 0.044891294091939926,
      "learning_rate": 0.0002474074074074074,
      "loss": 0.5809,
      "step": 348
    },
    {
      "epoch": 0.7742651136993899,
      "grad_norm": 0.04109985753893852,
      "learning_rate": 0.0002471604938271605,
      "loss": 0.5119,
      "step": 349
    },
    {
      "epoch": 0.776483638380477,
      "grad_norm": 0.07787656784057617,
      "learning_rate": 0.0002469135802469136,
      "loss": 0.769,
      "step": 350
    },
    {
      "epoch": 0.778702163061564,
      "grad_norm": 0.039786648005247116,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.4511,
      "step": 351
    },
    {
      "epoch": 0.7809206877426511,
      "grad_norm": 0.04101479798555374,
      "learning_rate": 0.0002464197530864197,
      "loss": 0.6061,
      "step": 352
    },
    {
      "epoch": 0.7831392124237382,
      "grad_norm": 0.04200151190161705,
      "learning_rate": 0.0002461728395061728,
      "loss": 0.5968,
      "step": 353
    },
    {
      "epoch": 0.7853577371048253,
      "grad_norm": 0.044166289269924164,
      "learning_rate": 0.0002459259259259259,
      "loss": 0.6053,
      "step": 354
    },
    {
      "epoch": 0.7875762617859123,
      "grad_norm": 0.059748973697423935,
      "learning_rate": 0.000245679012345679,
      "loss": 0.778,
      "step": 355
    },
    {
      "epoch": 0.7897947864669994,
      "grad_norm": 0.04654362052679062,
      "learning_rate": 0.00024543209876543205,
      "loss": 0.6711,
      "step": 356
    },
    {
      "epoch": 0.7920133111480865,
      "grad_norm": 0.041838426142930984,
      "learning_rate": 0.0002451851851851852,
      "loss": 0.647,
      "step": 357
    },
    {
      "epoch": 0.7942318358291736,
      "grad_norm": 0.04124513268470764,
      "learning_rate": 0.00024493827160493824,
      "loss": 0.5515,
      "step": 358
    },
    {
      "epoch": 0.7964503605102606,
      "grad_norm": 0.04266709089279175,
      "learning_rate": 0.00024469135802469133,
      "loss": 0.5691,
      "step": 359
    },
    {
      "epoch": 0.7986688851913477,
      "grad_norm": 0.06698063015937805,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.7121,
      "step": 360
    },
    {
      "epoch": 0.8008874098724348,
      "grad_norm": 0.047163620591163635,
      "learning_rate": 0.0002441975308641975,
      "loss": 0.5937,
      "step": 361
    },
    {
      "epoch": 0.8031059345535219,
      "grad_norm": 0.05831912159919739,
      "learning_rate": 0.00024395061728395062,
      "loss": 0.4818,
      "step": 362
    },
    {
      "epoch": 0.8053244592346089,
      "grad_norm": 0.036670587956905365,
      "learning_rate": 0.0002437037037037037,
      "loss": 0.5762,
      "step": 363
    },
    {
      "epoch": 0.807542983915696,
      "grad_norm": 0.04457823932170868,
      "learning_rate": 0.00024345679012345678,
      "loss": 0.7512,
      "step": 364
    },
    {
      "epoch": 0.8097615085967831,
      "grad_norm": 0.049701303243637085,
      "learning_rate": 0.00024320987654320985,
      "loss": 0.7709,
      "step": 365
    },
    {
      "epoch": 0.8119800332778702,
      "grad_norm": 0.04273749142885208,
      "learning_rate": 0.00024296296296296295,
      "loss": 0.4515,
      "step": 366
    },
    {
      "epoch": 0.8141985579589573,
      "grad_norm": 0.059366095811128616,
      "learning_rate": 0.00024271604938271601,
      "loss": 0.7633,
      "step": 367
    },
    {
      "epoch": 0.8164170826400444,
      "grad_norm": 0.04048094153404236,
      "learning_rate": 0.0002424691358024691,
      "loss": 0.588,
      "step": 368
    },
    {
      "epoch": 0.8186356073211315,
      "grad_norm": 0.07439504563808441,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.726,
      "step": 369
    },
    {
      "epoch": 0.8208541320022186,
      "grad_norm": 0.036472488194704056,
      "learning_rate": 0.00024197530864197527,
      "loss": 0.6412,
      "step": 370
    },
    {
      "epoch": 0.8230726566833056,
      "grad_norm": 0.03904552757740021,
      "learning_rate": 0.0002417283950617284,
      "loss": 0.5397,
      "step": 371
    },
    {
      "epoch": 0.8252911813643927,
      "grad_norm": 0.05665183812379837,
      "learning_rate": 0.00024148148148148146,
      "loss": 0.6317,
      "step": 372
    },
    {
      "epoch": 0.8275097060454798,
      "grad_norm": 0.04541141912341118,
      "learning_rate": 0.00024123456790123456,
      "loss": 0.5614,
      "step": 373
    },
    {
      "epoch": 0.8297282307265669,
      "grad_norm": 0.05993746966123581,
      "learning_rate": 0.00024098765432098763,
      "loss": 0.5686,
      "step": 374
    },
    {
      "epoch": 0.831946755407654,
      "grad_norm": 0.10583385825157166,
      "learning_rate": 0.00024074074074074072,
      "loss": 0.5224,
      "step": 375
    },
    {
      "epoch": 0.834165280088741,
      "grad_norm": 0.04051755741238594,
      "learning_rate": 0.0002404938271604938,
      "loss": 0.7036,
      "step": 376
    },
    {
      "epoch": 0.8363838047698281,
      "grad_norm": 0.05001147463917732,
      "learning_rate": 0.00024024691358024688,
      "loss": 0.6766,
      "step": 377
    },
    {
      "epoch": 0.8386023294509152,
      "grad_norm": 0.048899732530117035,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.8312,
      "step": 378
    },
    {
      "epoch": 0.8408208541320022,
      "grad_norm": 0.08577731996774673,
      "learning_rate": 0.00023975308641975305,
      "loss": 0.6153,
      "step": 379
    },
    {
      "epoch": 0.8430393788130893,
      "grad_norm": 0.06327125430107117,
      "learning_rate": 0.00023950617283950617,
      "loss": 0.5428,
      "step": 380
    },
    {
      "epoch": 0.8452579034941764,
      "grad_norm": 0.052416350692510605,
      "learning_rate": 0.00023925925925925924,
      "loss": 0.6005,
      "step": 381
    },
    {
      "epoch": 0.8474764281752635,
      "grad_norm": 0.05717407166957855,
      "learning_rate": 0.00023901234567901233,
      "loss": 0.5356,
      "step": 382
    },
    {
      "epoch": 0.8496949528563505,
      "grad_norm": 0.03670477867126465,
      "learning_rate": 0.0002387654320987654,
      "loss": 0.604,
      "step": 383
    },
    {
      "epoch": 0.8519134775374376,
      "grad_norm": 0.047979772090911865,
      "learning_rate": 0.0002385185185185185,
      "loss": 0.4505,
      "step": 384
    },
    {
      "epoch": 0.8541320022185247,
      "grad_norm": 0.04311596602201462,
      "learning_rate": 0.0002382716049382716,
      "loss": 0.5077,
      "step": 385
    },
    {
      "epoch": 0.8563505268996118,
      "grad_norm": 0.16468545794487,
      "learning_rate": 0.00023802469135802466,
      "loss": 0.5893,
      "step": 386
    },
    {
      "epoch": 0.8585690515806988,
      "grad_norm": 0.048438701778650284,
      "learning_rate": 0.00023777777777777775,
      "loss": 0.8592,
      "step": 387
    },
    {
      "epoch": 0.8607875762617859,
      "grad_norm": 0.03904375433921814,
      "learning_rate": 0.00023753086419753082,
      "loss": 0.5174,
      "step": 388
    },
    {
      "epoch": 0.863006100942873,
      "grad_norm": 0.05924297496676445,
      "learning_rate": 0.00023728395061728394,
      "loss": 0.564,
      "step": 389
    },
    {
      "epoch": 0.8652246256239601,
      "grad_norm": 0.04548504948616028,
      "learning_rate": 0.000237037037037037,
      "loss": 0.6231,
      "step": 390
    },
    {
      "epoch": 0.8674431503050472,
      "grad_norm": 0.04987689480185509,
      "learning_rate": 0.0002367901234567901,
      "loss": 0.615,
      "step": 391
    },
    {
      "epoch": 0.8696616749861342,
      "grad_norm": 0.0394514799118042,
      "learning_rate": 0.0002365432098765432,
      "loss": 0.5101,
      "step": 392
    },
    {
      "epoch": 0.8718801996672213,
      "grad_norm": 0.03759310394525528,
      "learning_rate": 0.00023629629629629627,
      "loss": 0.5624,
      "step": 393
    },
    {
      "epoch": 0.8740987243483084,
      "grad_norm": 0.03783238306641579,
      "learning_rate": 0.00023604938271604937,
      "loss": 0.547,
      "step": 394
    },
    {
      "epoch": 0.8763172490293955,
      "grad_norm": 0.04294927045702934,
      "learning_rate": 0.00023580246913580243,
      "loss": 0.552,
      "step": 395
    },
    {
      "epoch": 0.8785357737104825,
      "grad_norm": 0.03696984797716141,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.568,
      "step": 396
    },
    {
      "epoch": 0.8807542983915696,
      "grad_norm": 0.04781077429652214,
      "learning_rate": 0.0002353086419753086,
      "loss": 0.6392,
      "step": 397
    },
    {
      "epoch": 0.8829728230726567,
      "grad_norm": 0.039796147495508194,
      "learning_rate": 0.00023506172839506172,
      "loss": 0.6322,
      "step": 398
    },
    {
      "epoch": 0.8851913477537438,
      "grad_norm": 0.054580919444561005,
      "learning_rate": 0.0002348148148148148,
      "loss": 0.7177,
      "step": 399
    },
    {
      "epoch": 0.8874098724348308,
      "grad_norm": 0.04119761288166046,
      "learning_rate": 0.00023456790123456788,
      "loss": 0.6113,
      "step": 400
    },
    {
      "epoch": 0.8896283971159179,
      "grad_norm": 0.04393110051751137,
      "learning_rate": 0.00023432098765432098,
      "loss": 0.4985,
      "step": 401
    },
    {
      "epoch": 0.891846921797005,
      "grad_norm": 0.0478365495800972,
      "learning_rate": 0.00023407407407407405,
      "loss": 0.5744,
      "step": 402
    },
    {
      "epoch": 0.894065446478092,
      "grad_norm": 0.05157410725951195,
      "learning_rate": 0.00023382716049382714,
      "loss": 0.5337,
      "step": 403
    },
    {
      "epoch": 0.8962839711591791,
      "grad_norm": 0.05461892858147621,
      "learning_rate": 0.0002335802469135802,
      "loss": 0.5895,
      "step": 404
    },
    {
      "epoch": 0.8985024958402662,
      "grad_norm": 0.06308047473430634,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.6059,
      "step": 405
    },
    {
      "epoch": 0.9007210205213533,
      "grad_norm": 0.04055093601346016,
      "learning_rate": 0.00023308641975308637,
      "loss": 0.4034,
      "step": 406
    },
    {
      "epoch": 0.9029395452024404,
      "grad_norm": 0.04563959315419197,
      "learning_rate": 0.0002328395061728395,
      "loss": 0.5859,
      "step": 407
    },
    {
      "epoch": 0.9051580698835274,
      "grad_norm": 0.05770420655608177,
      "learning_rate": 0.0002325925925925926,
      "loss": 0.828,
      "step": 408
    },
    {
      "epoch": 0.9073765945646145,
      "grad_norm": 0.03846162185072899,
      "learning_rate": 0.00023234567901234566,
      "loss": 0.6612,
      "step": 409
    },
    {
      "epoch": 0.9095951192457016,
      "grad_norm": 0.06043208763003349,
      "learning_rate": 0.00023209876543209875,
      "loss": 0.6388,
      "step": 410
    },
    {
      "epoch": 0.9118136439267887,
      "grad_norm": 0.050817690789699554,
      "learning_rate": 0.00023185185185185182,
      "loss": 0.5459,
      "step": 411
    },
    {
      "epoch": 0.9140321686078757,
      "grad_norm": 0.05286123603582382,
      "learning_rate": 0.00023160493827160492,
      "loss": 0.6137,
      "step": 412
    },
    {
      "epoch": 0.9162506932889628,
      "grad_norm": 0.053144823759794235,
      "learning_rate": 0.00023135802469135798,
      "loss": 0.4968,
      "step": 413
    },
    {
      "epoch": 0.9184692179700499,
      "grad_norm": 0.050602059811353683,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.5937,
      "step": 414
    },
    {
      "epoch": 0.920687742651137,
      "grad_norm": 0.03789098933339119,
      "learning_rate": 0.0002308641975308642,
      "loss": 0.5253,
      "step": 415
    },
    {
      "epoch": 0.922906267332224,
      "grad_norm": 0.040765825659036636,
      "learning_rate": 0.00023061728395061727,
      "loss": 0.5798,
      "step": 416
    },
    {
      "epoch": 0.9251247920133111,
      "grad_norm": 0.06392791122198105,
      "learning_rate": 0.00023037037037037036,
      "loss": 0.5874,
      "step": 417
    },
    {
      "epoch": 0.9273433166943982,
      "grad_norm": 0.058316610753536224,
      "learning_rate": 0.00023012345679012343,
      "loss": 0.6716,
      "step": 418
    },
    {
      "epoch": 0.9295618413754853,
      "grad_norm": 0.05706488713622093,
      "learning_rate": 0.00022987654320987653,
      "loss": 0.6315,
      "step": 419
    },
    {
      "epoch": 0.9317803660565723,
      "grad_norm": 0.04155052453279495,
      "learning_rate": 0.0002296296296296296,
      "loss": 0.5528,
      "step": 420
    },
    {
      "epoch": 0.9339988907376594,
      "grad_norm": 0.07801178842782974,
      "learning_rate": 0.0002293827160493827,
      "loss": 0.6092,
      "step": 421
    },
    {
      "epoch": 0.9362174154187465,
      "grad_norm": 0.036215245723724365,
      "learning_rate": 0.00022913580246913579,
      "loss": 0.7033,
      "step": 422
    },
    {
      "epoch": 0.9384359400998337,
      "grad_norm": 0.06249186396598816,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.6497,
      "step": 423
    },
    {
      "epoch": 0.9406544647809207,
      "grad_norm": 0.04538951441645622,
      "learning_rate": 0.00022864197530864198,
      "loss": 0.6499,
      "step": 424
    },
    {
      "epoch": 0.9428729894620078,
      "grad_norm": 0.04153716191649437,
      "learning_rate": 0.00022839506172839504,
      "loss": 0.6556,
      "step": 425
    },
    {
      "epoch": 0.9450915141430949,
      "grad_norm": 0.05624442175030708,
      "learning_rate": 0.00022814814814814814,
      "loss": 0.694,
      "step": 426
    },
    {
      "epoch": 0.947310038824182,
      "grad_norm": 0.0400119349360466,
      "learning_rate": 0.0002279012345679012,
      "loss": 0.5746,
      "step": 427
    },
    {
      "epoch": 0.949528563505269,
      "grad_norm": 0.04542658478021622,
      "learning_rate": 0.0002276543209876543,
      "loss": 0.6452,
      "step": 428
    },
    {
      "epoch": 0.9517470881863561,
      "grad_norm": 0.04576200619339943,
      "learning_rate": 0.00022740740740740737,
      "loss": 0.6247,
      "step": 429
    },
    {
      "epoch": 0.9539656128674432,
      "grad_norm": 0.04086817428469658,
      "learning_rate": 0.00022716049382716047,
      "loss": 0.4884,
      "step": 430
    },
    {
      "epoch": 0.9561841375485303,
      "grad_norm": 0.045220617204904556,
      "learning_rate": 0.00022691358024691356,
      "loss": 0.6013,
      "step": 431
    },
    {
      "epoch": 0.9584026622296173,
      "grad_norm": 0.05225265398621559,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.6713,
      "step": 432
    },
    {
      "epoch": 0.9606211869107044,
      "grad_norm": 0.04428958520293236,
      "learning_rate": 0.00022641975308641975,
      "loss": 0.6559,
      "step": 433
    },
    {
      "epoch": 0.9628397115917915,
      "grad_norm": 0.04856949672102928,
      "learning_rate": 0.00022617283950617282,
      "loss": 0.74,
      "step": 434
    },
    {
      "epoch": 0.9650582362728786,
      "grad_norm": 0.0537244975566864,
      "learning_rate": 0.00022592592592592591,
      "loss": 0.6307,
      "step": 435
    },
    {
      "epoch": 0.9672767609539656,
      "grad_norm": 0.04043446481227875,
      "learning_rate": 0.00022567901234567898,
      "loss": 0.5983,
      "step": 436
    },
    {
      "epoch": 0.9694952856350527,
      "grad_norm": 0.03694366663694382,
      "learning_rate": 0.00022543209876543208,
      "loss": 0.5699,
      "step": 437
    },
    {
      "epoch": 0.9717138103161398,
      "grad_norm": 0.038097597658634186,
      "learning_rate": 0.00022518518518518517,
      "loss": 0.4814,
      "step": 438
    },
    {
      "epoch": 0.9739323349972269,
      "grad_norm": 0.03794325515627861,
      "learning_rate": 0.00022493827160493824,
      "loss": 0.5776,
      "step": 439
    },
    {
      "epoch": 0.9761508596783139,
      "grad_norm": 0.05676797404885292,
      "learning_rate": 0.00022469135802469134,
      "loss": 0.7681,
      "step": 440
    },
    {
      "epoch": 0.978369384359401,
      "grad_norm": 0.058536458760499954,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.5885,
      "step": 441
    },
    {
      "epoch": 0.9805879090404881,
      "grad_norm": 0.04586883261799812,
      "learning_rate": 0.00022419753086419753,
      "loss": 0.5402,
      "step": 442
    },
    {
      "epoch": 0.9828064337215752,
      "grad_norm": 0.04316874220967293,
      "learning_rate": 0.0002239506172839506,
      "loss": 0.7824,
      "step": 443
    },
    {
      "epoch": 0.9850249584026622,
      "grad_norm": 0.05695241317152977,
      "learning_rate": 0.0002237037037037037,
      "loss": 0.5238,
      "step": 444
    },
    {
      "epoch": 0.9872434830837493,
      "grad_norm": 0.03407418355345726,
      "learning_rate": 0.00022345679012345678,
      "loss": 0.5493,
      "step": 445
    },
    {
      "epoch": 0.9894620077648364,
      "grad_norm": 0.05841716378927231,
      "learning_rate": 0.00022320987654320985,
      "loss": 0.6694,
      "step": 446
    },
    {
      "epoch": 0.9916805324459235,
      "grad_norm": 0.044965095818042755,
      "learning_rate": 0.00022296296296296295,
      "loss": 0.5156,
      "step": 447
    },
    {
      "epoch": 0.9938990571270105,
      "grad_norm": 0.05165541172027588,
      "learning_rate": 0.00022271604938271602,
      "loss": 0.5728,
      "step": 448
    },
    {
      "epoch": 0.9961175818080976,
      "grad_norm": 0.056687381118535995,
      "learning_rate": 0.0002224691358024691,
      "loss": 0.6933,
      "step": 449
    },
    {
      "epoch": 0.9983361064891847,
      "grad_norm": 0.04889079928398132,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.6468,
      "step": 450
    },
    {
      "epoch": 1.0005546311702718,
      "grad_norm": 0.04568987712264061,
      "learning_rate": 0.0002219753086419753,
      "loss": 0.7504,
      "step": 451
    },
    {
      "epoch": 1.0027731558513588,
      "grad_norm": 0.04992784559726715,
      "learning_rate": 0.0002217283950617284,
      "loss": 0.6965,
      "step": 452
    },
    {
      "epoch": 1.004991680532446,
      "grad_norm": 0.04450535774230957,
      "learning_rate": 0.00022148148148148146,
      "loss": 0.578,
      "step": 453
    },
    {
      "epoch": 1.007210205213533,
      "grad_norm": 0.03635558485984802,
      "learning_rate": 0.00022123456790123456,
      "loss": 0.486,
      "step": 454
    },
    {
      "epoch": 1.00942872989462,
      "grad_norm": 0.04396234452724457,
      "learning_rate": 0.00022098765432098763,
      "loss": 0.4949,
      "step": 455
    },
    {
      "epoch": 1.0116472545757071,
      "grad_norm": 0.041818417608737946,
      "learning_rate": 0.00022074074074074072,
      "loss": 0.6724,
      "step": 456
    },
    {
      "epoch": 1.0138657792567942,
      "grad_norm": 0.050360262393951416,
      "learning_rate": 0.0002204938271604938,
      "loss": 0.5044,
      "step": 457
    },
    {
      "epoch": 1.0160843039378813,
      "grad_norm": 0.03422829881310463,
      "learning_rate": 0.00022024691358024689,
      "loss": 0.4858,
      "step": 458
    },
    {
      "epoch": 1.0183028286189684,
      "grad_norm": 0.03881603479385376,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.5566,
      "step": 459
    },
    {
      "epoch": 1.0205213533000554,
      "grad_norm": 0.03896480053663254,
      "learning_rate": 0.00021975308641975308,
      "loss": 0.4528,
      "step": 460
    },
    {
      "epoch": 1.0227398779811425,
      "grad_norm": 0.059975747019052505,
      "learning_rate": 0.00021950617283950617,
      "loss": 0.4984,
      "step": 461
    },
    {
      "epoch": 1.0249584026622296,
      "grad_norm": 0.046806659549474716,
      "learning_rate": 0.00021925925925925924,
      "loss": 0.6415,
      "step": 462
    },
    {
      "epoch": 1.0271769273433167,
      "grad_norm": 0.06901863217353821,
      "learning_rate": 0.00021901234567901233,
      "loss": 0.4785,
      "step": 463
    },
    {
      "epoch": 1.0293954520244037,
      "grad_norm": 0.043647654354572296,
      "learning_rate": 0.0002187654320987654,
      "loss": 0.5258,
      "step": 464
    },
    {
      "epoch": 1.0316139767054908,
      "grad_norm": 0.0463489331305027,
      "learning_rate": 0.0002185185185185185,
      "loss": 0.5553,
      "step": 465
    },
    {
      "epoch": 1.033832501386578,
      "grad_norm": 0.04219133034348488,
      "learning_rate": 0.00021827160493827157,
      "loss": 0.598,
      "step": 466
    },
    {
      "epoch": 1.036051026067665,
      "grad_norm": 0.057779956609010696,
      "learning_rate": 0.00021802469135802466,
      "loss": 0.653,
      "step": 467
    },
    {
      "epoch": 1.038269550748752,
      "grad_norm": 0.05019902065396309,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.732,
      "step": 468
    },
    {
      "epoch": 1.0404880754298391,
      "grad_norm": 0.06764944642782211,
      "learning_rate": 0.00021753086419753085,
      "loss": 0.5651,
      "step": 469
    },
    {
      "epoch": 1.0427066001109262,
      "grad_norm": 0.05941182002425194,
      "learning_rate": 0.00021728395061728395,
      "loss": 0.5354,
      "step": 470
    },
    {
      "epoch": 1.0449251247920133,
      "grad_norm": 0.04143821448087692,
      "learning_rate": 0.00021703703703703701,
      "loss": 0.5625,
      "step": 471
    },
    {
      "epoch": 1.0471436494731003,
      "grad_norm": 0.04522986337542534,
      "learning_rate": 0.0002167901234567901,
      "loss": 0.6766,
      "step": 472
    },
    {
      "epoch": 1.0493621741541874,
      "grad_norm": 0.042082928121089935,
      "learning_rate": 0.00021654320987654318,
      "loss": 0.585,
      "step": 473
    },
    {
      "epoch": 1.0515806988352745,
      "grad_norm": 0.033250827342271805,
      "learning_rate": 0.00021629629629629627,
      "loss": 0.6303,
      "step": 474
    },
    {
      "epoch": 1.0537992235163616,
      "grad_norm": 0.06376510113477707,
      "learning_rate": 0.00021604938271604937,
      "loss": 0.5868,
      "step": 475
    },
    {
      "epoch": 1.0560177481974486,
      "grad_norm": 0.05008485168218613,
      "learning_rate": 0.00021580246913580244,
      "loss": 0.6205,
      "step": 476
    },
    {
      "epoch": 1.0582362728785357,
      "grad_norm": 0.043864622712135315,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.4122,
      "step": 477
    },
    {
      "epoch": 1.0604547975596228,
      "grad_norm": 0.03856130316853523,
      "learning_rate": 0.00021530864197530863,
      "loss": 0.4816,
      "step": 478
    },
    {
      "epoch": 1.0626733222407099,
      "grad_norm": 0.04481562599539757,
      "learning_rate": 0.00021506172839506172,
      "loss": 0.5345,
      "step": 479
    },
    {
      "epoch": 1.064891846921797,
      "grad_norm": 0.054971884936094284,
      "learning_rate": 0.0002148148148148148,
      "loss": 0.7347,
      "step": 480
    },
    {
      "epoch": 1.067110371602884,
      "grad_norm": 0.0546475350856781,
      "learning_rate": 0.00021456790123456788,
      "loss": 0.5436,
      "step": 481
    },
    {
      "epoch": 1.069328896283971,
      "grad_norm": 0.05842655897140503,
      "learning_rate": 0.00021432098765432095,
      "loss": 0.6596,
      "step": 482
    },
    {
      "epoch": 1.0715474209650582,
      "grad_norm": 0.05079928785562515,
      "learning_rate": 0.00021407407407407405,
      "loss": 0.7797,
      "step": 483
    },
    {
      "epoch": 1.0737659456461452,
      "grad_norm": 0.0608680434525013,
      "learning_rate": 0.00021382716049382714,
      "loss": 0.6235,
      "step": 484
    },
    {
      "epoch": 1.0759844703272323,
      "grad_norm": 0.048657506704330444,
      "learning_rate": 0.0002135802469135802,
      "loss": 0.4652,
      "step": 485
    },
    {
      "epoch": 1.0782029950083194,
      "grad_norm": 0.04425787553191185,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.5398,
      "step": 486
    },
    {
      "epoch": 1.0804215196894065,
      "grad_norm": 0.053888365626335144,
      "learning_rate": 0.0002130864197530864,
      "loss": 0.4704,
      "step": 487
    },
    {
      "epoch": 1.0826400443704935,
      "grad_norm": 0.04263632372021675,
      "learning_rate": 0.0002128395061728395,
      "loss": 0.6394,
      "step": 488
    },
    {
      "epoch": 1.0848585690515806,
      "grad_norm": 0.07079622894525528,
      "learning_rate": 0.00021259259259259256,
      "loss": 0.8555,
      "step": 489
    },
    {
      "epoch": 1.0870770937326677,
      "grad_norm": 0.04188642278313637,
      "learning_rate": 0.00021234567901234566,
      "loss": 0.7367,
      "step": 490
    },
    {
      "epoch": 1.0892956184137548,
      "grad_norm": 0.08358854800462723,
      "learning_rate": 0.00021209876543209875,
      "loss": 0.815,
      "step": 491
    },
    {
      "epoch": 1.0915141430948418,
      "grad_norm": 0.028964629396796227,
      "learning_rate": 0.00021185185185185182,
      "loss": 0.5434,
      "step": 492
    },
    {
      "epoch": 1.093732667775929,
      "grad_norm": 0.054062794893980026,
      "learning_rate": 0.00021160493827160492,
      "loss": 0.7186,
      "step": 493
    },
    {
      "epoch": 1.095951192457016,
      "grad_norm": 0.0529790073633194,
      "learning_rate": 0.00021135802469135799,
      "loss": 0.6293,
      "step": 494
    },
    {
      "epoch": 1.098169717138103,
      "grad_norm": 0.05028403922915459,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.4765,
      "step": 495
    },
    {
      "epoch": 1.1003882418191901,
      "grad_norm": 0.04172990471124649,
      "learning_rate": 0.00021086419753086418,
      "loss": 0.6764,
      "step": 496
    },
    {
      "epoch": 1.1026067665002772,
      "grad_norm": 0.036900538951158524,
      "learning_rate": 0.00021061728395061727,
      "loss": 0.6301,
      "step": 497
    },
    {
      "epoch": 1.1048252911813643,
      "grad_norm": 0.03885071352124214,
      "learning_rate": 0.00021037037037037037,
      "loss": 0.5576,
      "step": 498
    },
    {
      "epoch": 1.1070438158624514,
      "grad_norm": 0.05007636919617653,
      "learning_rate": 0.00021012345679012343,
      "loss": 0.6082,
      "step": 499
    },
    {
      "epoch": 1.1092623405435384,
      "grad_norm": 0.042398322373628616,
      "learning_rate": 0.00020987654320987653,
      "loss": 0.6316,
      "step": 500
    },
    {
      "epoch": 1.1114808652246255,
      "grad_norm": 0.04685323312878609,
      "learning_rate": 0.0002096296296296296,
      "loss": 0.5387,
      "step": 501
    },
    {
      "epoch": 1.1136993899057126,
      "grad_norm": 0.04394229128956795,
      "learning_rate": 0.0002093827160493827,
      "loss": 0.5368,
      "step": 502
    },
    {
      "epoch": 1.1159179145867997,
      "grad_norm": 0.0373038612306118,
      "learning_rate": 0.00020913580246913576,
      "loss": 0.6306,
      "step": 503
    },
    {
      "epoch": 1.1181364392678868,
      "grad_norm": 0.07625854015350342,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.6357,
      "step": 504
    },
    {
      "epoch": 1.1203549639489738,
      "grad_norm": 0.03662346675992012,
      "learning_rate": 0.00020864197530864198,
      "loss": 0.5424,
      "step": 505
    },
    {
      "epoch": 1.122573488630061,
      "grad_norm": 0.04083077609539032,
      "learning_rate": 0.00020839506172839505,
      "loss": 0.5351,
      "step": 506
    },
    {
      "epoch": 1.124792013311148,
      "grad_norm": 0.04142534360289574,
      "learning_rate": 0.00020814814814814814,
      "loss": 0.6298,
      "step": 507
    },
    {
      "epoch": 1.127010537992235,
      "grad_norm": 0.05522964894771576,
      "learning_rate": 0.0002079012345679012,
      "loss": 0.5492,
      "step": 508
    },
    {
      "epoch": 1.1292290626733221,
      "grad_norm": 0.04608926549553871,
      "learning_rate": 0.0002076543209876543,
      "loss": 0.4944,
      "step": 509
    },
    {
      "epoch": 1.1314475873544092,
      "grad_norm": 0.04951908811926842,
      "learning_rate": 0.00020740740740740737,
      "loss": 0.6535,
      "step": 510
    },
    {
      "epoch": 1.1336661120354963,
      "grad_norm": 0.040972691029310226,
      "learning_rate": 0.00020716049382716047,
      "loss": 0.6798,
      "step": 511
    },
    {
      "epoch": 1.1358846367165834,
      "grad_norm": 0.04819873347878456,
      "learning_rate": 0.00020691358024691354,
      "loss": 0.7515,
      "step": 512
    },
    {
      "epoch": 1.1381031613976704,
      "grad_norm": 0.05089031904935837,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.6705,
      "step": 513
    },
    {
      "epoch": 1.1403216860787575,
      "grad_norm": 0.04260292276740074,
      "learning_rate": 0.00020641975308641975,
      "loss": 0.4648,
      "step": 514
    },
    {
      "epoch": 1.1425402107598448,
      "grad_norm": 0.061396993696689606,
      "learning_rate": 0.00020617283950617282,
      "loss": 0.6647,
      "step": 515
    },
    {
      "epoch": 1.1447587354409319,
      "grad_norm": 0.03814017400145531,
      "learning_rate": 0.00020592592592592592,
      "loss": 0.6451,
      "step": 516
    },
    {
      "epoch": 1.146977260122019,
      "grad_norm": 0.04623452574014664,
      "learning_rate": 0.00020567901234567898,
      "loss": 0.6272,
      "step": 517
    },
    {
      "epoch": 1.149195784803106,
      "grad_norm": 0.0629858449101448,
      "learning_rate": 0.00020543209876543208,
      "loss": 0.6566,
      "step": 518
    },
    {
      "epoch": 1.151414309484193,
      "grad_norm": 0.060075730085372925,
      "learning_rate": 0.00020518518518518515,
      "loss": 0.7003,
      "step": 519
    },
    {
      "epoch": 1.1536328341652802,
      "grad_norm": 0.04917770251631737,
      "learning_rate": 0.00020493827160493824,
      "loss": 0.5165,
      "step": 520
    },
    {
      "epoch": 1.1558513588463672,
      "grad_norm": 0.03316838666796684,
      "learning_rate": 0.00020469135802469136,
      "loss": 0.5739,
      "step": 521
    },
    {
      "epoch": 1.1580698835274543,
      "grad_norm": 0.034732792526483536,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.6033,
      "step": 522
    },
    {
      "epoch": 1.1602884082085414,
      "grad_norm": 0.03628681227564812,
      "learning_rate": 0.00020419753086419753,
      "loss": 0.5684,
      "step": 523
    },
    {
      "epoch": 1.1625069328896285,
      "grad_norm": 0.04794244095683098,
      "learning_rate": 0.0002039506172839506,
      "loss": 0.5298,
      "step": 524
    },
    {
      "epoch": 1.1647254575707155,
      "grad_norm": 0.051314108073711395,
      "learning_rate": 0.0002037037037037037,
      "loss": 0.6718,
      "step": 525
    },
    {
      "epoch": 1.1669439822518026,
      "grad_norm": 0.042436808347702026,
      "learning_rate": 0.00020345679012345676,
      "loss": 0.6709,
      "step": 526
    },
    {
      "epoch": 1.1691625069328897,
      "grad_norm": 0.04300607368350029,
      "learning_rate": 0.00020320987654320985,
      "loss": 0.4921,
      "step": 527
    },
    {
      "epoch": 1.1713810316139768,
      "grad_norm": 0.039465904235839844,
      "learning_rate": 0.00020296296296296295,
      "loss": 0.5441,
      "step": 528
    },
    {
      "epoch": 1.1735995562950639,
      "grad_norm": 0.04602843150496483,
      "learning_rate": 0.00020271604938271602,
      "loss": 0.7194,
      "step": 529
    },
    {
      "epoch": 1.175818080976151,
      "grad_norm": 0.03710653632879257,
      "learning_rate": 0.00020246913580246914,
      "loss": 0.5312,
      "step": 530
    },
    {
      "epoch": 1.178036605657238,
      "grad_norm": 0.07612698525190353,
      "learning_rate": 0.0002022222222222222,
      "loss": 0.6535,
      "step": 531
    },
    {
      "epoch": 1.180255130338325,
      "grad_norm": 0.04423061013221741,
      "learning_rate": 0.0002019753086419753,
      "loss": 0.5728,
      "step": 532
    },
    {
      "epoch": 1.1824736550194122,
      "grad_norm": 0.08643504977226257,
      "learning_rate": 0.00020172839506172837,
      "loss": 0.6452,
      "step": 533
    },
    {
      "epoch": 1.1846921797004992,
      "grad_norm": 0.050138920545578,
      "learning_rate": 0.00020148148148148147,
      "loss": 0.4441,
      "step": 534
    },
    {
      "epoch": 1.1869107043815863,
      "grad_norm": 0.03691035881638527,
      "learning_rate": 0.00020123456790123456,
      "loss": 0.5798,
      "step": 535
    },
    {
      "epoch": 1.1891292290626734,
      "grad_norm": 0.05790675804018974,
      "learning_rate": 0.00020098765432098763,
      "loss": 0.5524,
      "step": 536
    },
    {
      "epoch": 1.1913477537437605,
      "grad_norm": 0.04946612939238548,
      "learning_rate": 0.00020074074074074072,
      "loss": 0.5834,
      "step": 537
    },
    {
      "epoch": 1.1935662784248475,
      "grad_norm": 0.05146505683660507,
      "learning_rate": 0.0002004938271604938,
      "loss": 0.6409,
      "step": 538
    },
    {
      "epoch": 1.1957848031059346,
      "grad_norm": 0.039599910378456116,
      "learning_rate": 0.00020024691358024691,
      "loss": 0.5249,
      "step": 539
    },
    {
      "epoch": 1.1980033277870217,
      "grad_norm": 0.05941521376371384,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.5243,
      "step": 540
    },
    {
      "epoch": 1.2002218524681088,
      "grad_norm": 0.037457916885614395,
      "learning_rate": 0.00019975308641975308,
      "loss": 0.5659,
      "step": 541
    },
    {
      "epoch": 1.2024403771491958,
      "grad_norm": 0.051353029906749725,
      "learning_rate": 0.00019950617283950615,
      "loss": 0.5224,
      "step": 542
    },
    {
      "epoch": 1.204658901830283,
      "grad_norm": 0.04103579744696617,
      "learning_rate": 0.00019925925925925924,
      "loss": 0.5999,
      "step": 543
    },
    {
      "epoch": 1.20687742651137,
      "grad_norm": 0.040466394275426865,
      "learning_rate": 0.00019901234567901234,
      "loss": 0.481,
      "step": 544
    },
    {
      "epoch": 1.209095951192457,
      "grad_norm": 0.044617991894483566,
      "learning_rate": 0.0001987654320987654,
      "loss": 0.5211,
      "step": 545
    },
    {
      "epoch": 1.2113144758735441,
      "grad_norm": 0.043850068002939224,
      "learning_rate": 0.0001985185185185185,
      "loss": 0.4529,
      "step": 546
    },
    {
      "epoch": 1.2135330005546312,
      "grad_norm": 0.03748058155179024,
      "learning_rate": 0.00019827160493827157,
      "loss": 0.5897,
      "step": 547
    },
    {
      "epoch": 1.2157515252357183,
      "grad_norm": 0.059728819876909256,
      "learning_rate": 0.0001980246913580247,
      "loss": 0.5454,
      "step": 548
    },
    {
      "epoch": 1.2179700499168054,
      "grad_norm": 0.04588298499584198,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.7025,
      "step": 549
    },
    {
      "epoch": 1.2201885745978924,
      "grad_norm": 0.055026307702064514,
      "learning_rate": 0.00019753086419753085,
      "loss": 0.7149,
      "step": 550
    },
    {
      "epoch": 1.2224070992789795,
      "grad_norm": 0.04991525784134865,
      "learning_rate": 0.00019728395061728395,
      "loss": 0.6624,
      "step": 551
    },
    {
      "epoch": 1.2246256239600666,
      "grad_norm": 0.06671017408370972,
      "learning_rate": 0.00019703703703703702,
      "loss": 0.6075,
      "step": 552
    },
    {
      "epoch": 1.2268441486411537,
      "grad_norm": 0.035714324563741684,
      "learning_rate": 0.0001967901234567901,
      "loss": 0.5592,
      "step": 553
    },
    {
      "epoch": 1.2290626733222407,
      "grad_norm": 0.043823350220918655,
      "learning_rate": 0.00019654320987654318,
      "loss": 0.589,
      "step": 554
    },
    {
      "epoch": 1.2312811980033278,
      "grad_norm": 0.039831701666116714,
      "learning_rate": 0.00019629629629629627,
      "loss": 0.6548,
      "step": 555
    },
    {
      "epoch": 1.2334997226844149,
      "grad_norm": 0.050841763615608215,
      "learning_rate": 0.00019604938271604934,
      "loss": 0.6506,
      "step": 556
    },
    {
      "epoch": 1.235718247365502,
      "grad_norm": 0.04285004734992981,
      "learning_rate": 0.00019580246913580246,
      "loss": 0.5683,
      "step": 557
    },
    {
      "epoch": 1.237936772046589,
      "grad_norm": 0.042617760598659515,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.5415,
      "step": 558
    },
    {
      "epoch": 1.240155296727676,
      "grad_norm": 0.04030551761388779,
      "learning_rate": 0.00019530864197530863,
      "loss": 0.5601,
      "step": 559
    },
    {
      "epoch": 1.2423738214087632,
      "grad_norm": 0.040038153529167175,
      "learning_rate": 0.00019506172839506172,
      "loss": 0.5637,
      "step": 560
    },
    {
      "epoch": 1.2445923460898503,
      "grad_norm": 0.048313818871974945,
      "learning_rate": 0.0001948148148148148,
      "loss": 0.5242,
      "step": 561
    },
    {
      "epoch": 1.2468108707709373,
      "grad_norm": 0.060618359595537186,
      "learning_rate": 0.00019456790123456789,
      "loss": 0.6058,
      "step": 562
    },
    {
      "epoch": 1.2490293954520244,
      "grad_norm": 0.044483255594968796,
      "learning_rate": 0.00019432098765432095,
      "loss": 0.7094,
      "step": 563
    },
    {
      "epoch": 1.2512479201331115,
      "grad_norm": 0.044360872358083725,
      "learning_rate": 0.00019407407407407405,
      "loss": 0.6495,
      "step": 564
    },
    {
      "epoch": 1.2534664448141986,
      "grad_norm": 0.03786607086658478,
      "learning_rate": 0.00019382716049382712,
      "loss": 0.5441,
      "step": 565
    },
    {
      "epoch": 1.2556849694952856,
      "grad_norm": 0.05588468536734581,
      "learning_rate": 0.00019358024691358024,
      "loss": 0.7255,
      "step": 566
    },
    {
      "epoch": 1.2579034941763727,
      "grad_norm": 0.03696209937334061,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.6245,
      "step": 567
    },
    {
      "epoch": 1.2601220188574598,
      "grad_norm": 0.03493791073560715,
      "learning_rate": 0.0001930864197530864,
      "loss": 0.6183,
      "step": 568
    },
    {
      "epoch": 1.2623405435385469,
      "grad_norm": 0.027954673394560814,
      "learning_rate": 0.0001928395061728395,
      "loss": 0.5237,
      "step": 569
    },
    {
      "epoch": 1.264559068219634,
      "grad_norm": 0.03779634088277817,
      "learning_rate": 0.00019259259259259257,
      "loss": 0.5183,
      "step": 570
    },
    {
      "epoch": 1.266777592900721,
      "grad_norm": 0.040785398334264755,
      "learning_rate": 0.00019234567901234566,
      "loss": 0.5764,
      "step": 571
    },
    {
      "epoch": 1.268996117581808,
      "grad_norm": 0.05332619324326515,
      "learning_rate": 0.00019209876543209873,
      "loss": 0.6473,
      "step": 572
    },
    {
      "epoch": 1.2712146422628952,
      "grad_norm": 0.03947728872299194,
      "learning_rate": 0.00019185185185185182,
      "loss": 0.6481,
      "step": 573
    },
    {
      "epoch": 1.2734331669439822,
      "grad_norm": 0.050838902592659,
      "learning_rate": 0.00019160493827160495,
      "loss": 0.5854,
      "step": 574
    },
    {
      "epoch": 1.2756516916250693,
      "grad_norm": 0.05824773386120796,
      "learning_rate": 0.00019135802469135801,
      "loss": 0.7423,
      "step": 575
    },
    {
      "epoch": 1.2778702163061564,
      "grad_norm": 0.034553222358226776,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.5346,
      "step": 576
    },
    {
      "epoch": 1.2800887409872435,
      "grad_norm": 0.05219268426299095,
      "learning_rate": 0.00019086419753086418,
      "loss": 0.627,
      "step": 577
    },
    {
      "epoch": 1.2823072656683305,
      "grad_norm": 0.03489074856042862,
      "learning_rate": 0.00019061728395061727,
      "loss": 0.5986,
      "step": 578
    },
    {
      "epoch": 1.2845257903494176,
      "grad_norm": 0.030668819323182106,
      "learning_rate": 0.00019037037037037034,
      "loss": 0.4967,
      "step": 579
    },
    {
      "epoch": 1.2867443150305047,
      "grad_norm": 0.06149688735604286,
      "learning_rate": 0.00019012345679012344,
      "loss": 0.7745,
      "step": 580
    },
    {
      "epoch": 1.2889628397115918,
      "grad_norm": 0.03323828801512718,
      "learning_rate": 0.00018987654320987653,
      "loss": 0.7236,
      "step": 581
    },
    {
      "epoch": 1.2911813643926788,
      "grad_norm": 0.038051821291446686,
      "learning_rate": 0.0001896296296296296,
      "loss": 0.5343,
      "step": 582
    },
    {
      "epoch": 1.293399889073766,
      "grad_norm": 0.05149319767951965,
      "learning_rate": 0.00018938271604938272,
      "loss": 0.4918,
      "step": 583
    },
    {
      "epoch": 1.295618413754853,
      "grad_norm": 0.04399671405553818,
      "learning_rate": 0.0001891358024691358,
      "loss": 0.6037,
      "step": 584
    },
    {
      "epoch": 1.29783693843594,
      "grad_norm": 0.04472484067082405,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.6743,
      "step": 585
    },
    {
      "epoch": 1.3000554631170271,
      "grad_norm": 0.041158225387334824,
      "learning_rate": 0.00018864197530864195,
      "loss": 0.5142,
      "step": 586
    },
    {
      "epoch": 1.3022739877981142,
      "grad_norm": 0.04662076011300087,
      "learning_rate": 0.00018839506172839505,
      "loss": 0.4853,
      "step": 587
    },
    {
      "epoch": 1.3044925124792013,
      "grad_norm": 0.04465070366859436,
      "learning_rate": 0.00018814814814814814,
      "loss": 0.6253,
      "step": 588
    },
    {
      "epoch": 1.3067110371602884,
      "grad_norm": 0.04635493457317352,
      "learning_rate": 0.0001879012345679012,
      "loss": 0.6245,
      "step": 589
    },
    {
      "epoch": 1.3089295618413754,
      "grad_norm": 0.04973436892032623,
      "learning_rate": 0.0001876543209876543,
      "loss": 0.5926,
      "step": 590
    },
    {
      "epoch": 1.3111480865224625,
      "grad_norm": 0.03652235120534897,
      "learning_rate": 0.00018740740740740737,
      "loss": 0.4356,
      "step": 591
    },
    {
      "epoch": 1.3133666112035496,
      "grad_norm": 0.05986592173576355,
      "learning_rate": 0.0001871604938271605,
      "loss": 0.7122,
      "step": 592
    },
    {
      "epoch": 1.3155851358846367,
      "grad_norm": 0.035313576459884644,
      "learning_rate": 0.00018691358024691356,
      "loss": 0.5319,
      "step": 593
    },
    {
      "epoch": 1.3178036605657237,
      "grad_norm": 0.04377858340740204,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.7456,
      "step": 594
    },
    {
      "epoch": 1.3200221852468108,
      "grad_norm": 0.040692467242479324,
      "learning_rate": 0.00018641975308641973,
      "loss": 0.8123,
      "step": 595
    },
    {
      "epoch": 1.3222407099278979,
      "grad_norm": 0.039379578083753586,
      "learning_rate": 0.00018617283950617282,
      "loss": 0.4937,
      "step": 596
    },
    {
      "epoch": 1.324459234608985,
      "grad_norm": 0.04515698179602623,
      "learning_rate": 0.00018592592592592592,
      "loss": 0.4948,
      "step": 597
    },
    {
      "epoch": 1.326677759290072,
      "grad_norm": 0.05209259316325188,
      "learning_rate": 0.00018567901234567899,
      "loss": 0.5918,
      "step": 598
    },
    {
      "epoch": 1.328896283971159,
      "grad_norm": 0.04363837093114853,
      "learning_rate": 0.00018543209876543208,
      "loss": 0.406,
      "step": 599
    },
    {
      "epoch": 1.3311148086522462,
      "grad_norm": 0.03892426937818527,
      "learning_rate": 0.00018518518518518515,
      "loss": 0.6311,
      "step": 600
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.04435485973954201,
      "learning_rate": 0.00018493827160493827,
      "loss": 0.6846,
      "step": 601
    },
    {
      "epoch": 1.3355518580144203,
      "grad_norm": 0.05131830275058746,
      "learning_rate": 0.00018469135802469134,
      "loss": 0.682,
      "step": 602
    },
    {
      "epoch": 1.3377703826955074,
      "grad_norm": 0.0355299711227417,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.5407,
      "step": 603
    },
    {
      "epoch": 1.3399889073765945,
      "grad_norm": 0.046515557914972305,
      "learning_rate": 0.00018419753086419753,
      "loss": 0.6012,
      "step": 604
    },
    {
      "epoch": 1.3422074320576816,
      "grad_norm": 0.039434123784303665,
      "learning_rate": 0.0001839506172839506,
      "loss": 0.6043,
      "step": 605
    },
    {
      "epoch": 1.3444259567387689,
      "grad_norm": 0.050879448652267456,
      "learning_rate": 0.0001837037037037037,
      "loss": 0.7206,
      "step": 606
    },
    {
      "epoch": 1.346644481419856,
      "grad_norm": 0.05093839392066002,
      "learning_rate": 0.00018345679012345676,
      "loss": 0.7076,
      "step": 607
    },
    {
      "epoch": 1.348863006100943,
      "grad_norm": 0.04807393252849579,
      "learning_rate": 0.00018320987654320986,
      "loss": 0.6385,
      "step": 608
    },
    {
      "epoch": 1.35108153078203,
      "grad_norm": 0.038658760488033295,
      "learning_rate": 0.00018296296296296292,
      "loss": 0.7182,
      "step": 609
    },
    {
      "epoch": 1.3533000554631172,
      "grad_norm": 0.04393618553876877,
      "learning_rate": 0.00018271604938271605,
      "loss": 0.5719,
      "step": 610
    },
    {
      "epoch": 1.3555185801442042,
      "grad_norm": 0.03658636286854744,
      "learning_rate": 0.00018246913580246914,
      "loss": 0.4301,
      "step": 611
    },
    {
      "epoch": 1.3577371048252913,
      "grad_norm": 0.05456067994236946,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.589,
      "step": 612
    },
    {
      "epoch": 1.3599556295063784,
      "grad_norm": 0.045375995337963104,
      "learning_rate": 0.0001819753086419753,
      "loss": 0.6189,
      "step": 613
    },
    {
      "epoch": 1.3621741541874655,
      "grad_norm": 0.04385276511311531,
      "learning_rate": 0.00018172839506172837,
      "loss": 0.5971,
      "step": 614
    },
    {
      "epoch": 1.3643926788685525,
      "grad_norm": 0.037793878465890884,
      "learning_rate": 0.00018148148148148147,
      "loss": 0.5333,
      "step": 615
    },
    {
      "epoch": 1.3666112035496396,
      "grad_norm": 0.03864775598049164,
      "learning_rate": 0.00018123456790123454,
      "loss": 0.5136,
      "step": 616
    },
    {
      "epoch": 1.3688297282307267,
      "grad_norm": 0.052298009395599365,
      "learning_rate": 0.00018098765432098763,
      "loss": 0.7423,
      "step": 617
    },
    {
      "epoch": 1.3710482529118138,
      "grad_norm": 0.03520698845386505,
      "learning_rate": 0.00018074074074074075,
      "loss": 0.5954,
      "step": 618
    },
    {
      "epoch": 1.3732667775929008,
      "grad_norm": 0.045778244733810425,
      "learning_rate": 0.00018049382716049382,
      "loss": 0.5877,
      "step": 619
    },
    {
      "epoch": 1.375485302273988,
      "grad_norm": 0.05923708528280258,
      "learning_rate": 0.00018024691358024692,
      "loss": 0.5337,
      "step": 620
    },
    {
      "epoch": 1.377703826955075,
      "grad_norm": 0.036904435604810715,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.7138,
      "step": 621
    },
    {
      "epoch": 1.379922351636162,
      "grad_norm": 0.03553173318505287,
      "learning_rate": 0.00017975308641975308,
      "loss": 0.4475,
      "step": 622
    },
    {
      "epoch": 1.3821408763172491,
      "grad_norm": 0.05562874302268028,
      "learning_rate": 0.00017950617283950615,
      "loss": 0.6908,
      "step": 623
    },
    {
      "epoch": 1.3843594009983362,
      "grad_norm": 0.038246043026447296,
      "learning_rate": 0.00017925925925925924,
      "loss": 0.579,
      "step": 624
    },
    {
      "epoch": 1.3865779256794233,
      "grad_norm": 0.045788925141096115,
      "learning_rate": 0.0001790123456790123,
      "loss": 0.5753,
      "step": 625
    },
    {
      "epoch": 1.3887964503605104,
      "grad_norm": 0.03661423549056053,
      "learning_rate": 0.0001787654320987654,
      "loss": 0.4231,
      "step": 626
    },
    {
      "epoch": 1.3910149750415974,
      "grad_norm": 0.044681768864393234,
      "learning_rate": 0.00017851851851851853,
      "loss": 0.5674,
      "step": 627
    },
    {
      "epoch": 1.3932334997226845,
      "grad_norm": 0.04418782889842987,
      "learning_rate": 0.0001782716049382716,
      "loss": 0.5064,
      "step": 628
    },
    {
      "epoch": 1.3954520244037716,
      "grad_norm": 0.04721570014953613,
      "learning_rate": 0.0001780246913580247,
      "loss": 0.5658,
      "step": 629
    },
    {
      "epoch": 1.3976705490848587,
      "grad_norm": 0.038281917572021484,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.5573,
      "step": 630
    },
    {
      "epoch": 1.3998890737659457,
      "grad_norm": 0.06680215150117874,
      "learning_rate": 0.00017753086419753085,
      "loss": 0.5722,
      "step": 631
    },
    {
      "epoch": 1.4021075984470328,
      "grad_norm": 0.037746552377939224,
      "learning_rate": 0.00017728395061728392,
      "loss": 0.525,
      "step": 632
    },
    {
      "epoch": 1.4043261231281199,
      "grad_norm": 0.03642165660858154,
      "learning_rate": 0.00017703703703703702,
      "loss": 0.5966,
      "step": 633
    },
    {
      "epoch": 1.406544647809207,
      "grad_norm": 0.05030370503664017,
      "learning_rate": 0.0001767901234567901,
      "loss": 0.6043,
      "step": 634
    },
    {
      "epoch": 1.408763172490294,
      "grad_norm": 0.047312166541814804,
      "learning_rate": 0.00017654320987654318,
      "loss": 0.5284,
      "step": 635
    },
    {
      "epoch": 1.410981697171381,
      "grad_norm": 0.06951280683279037,
      "learning_rate": 0.0001762962962962963,
      "loss": 0.5785,
      "step": 636
    },
    {
      "epoch": 1.4132002218524682,
      "grad_norm": 0.038127876818180084,
      "learning_rate": 0.00017604938271604937,
      "loss": 0.5491,
      "step": 637
    },
    {
      "epoch": 1.4154187465335553,
      "grad_norm": 0.0668160691857338,
      "learning_rate": 0.00017580246913580247,
      "loss": 0.6391,
      "step": 638
    },
    {
      "epoch": 1.4176372712146423,
      "grad_norm": 0.06257199496030807,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.6514,
      "step": 639
    },
    {
      "epoch": 1.4198557958957294,
      "grad_norm": 0.04113186523318291,
      "learning_rate": 0.00017530864197530863,
      "loss": 0.5541,
      "step": 640
    },
    {
      "epoch": 1.4220743205768165,
      "grad_norm": 0.05734118074178696,
      "learning_rate": 0.00017506172839506172,
      "loss": 0.5599,
      "step": 641
    },
    {
      "epoch": 1.4242928452579036,
      "grad_norm": 0.035989902913570404,
      "learning_rate": 0.0001748148148148148,
      "loss": 0.5127,
      "step": 642
    },
    {
      "epoch": 1.4265113699389906,
      "grad_norm": 0.04810425639152527,
      "learning_rate": 0.0001745679012345679,
      "loss": 0.4757,
      "step": 643
    },
    {
      "epoch": 1.4287298946200777,
      "grad_norm": 0.06294404715299606,
      "learning_rate": 0.00017432098765432096,
      "loss": 0.7072,
      "step": 644
    },
    {
      "epoch": 1.4309484193011648,
      "grad_norm": 0.036472853273153305,
      "learning_rate": 0.00017407407407407408,
      "loss": 0.5292,
      "step": 645
    },
    {
      "epoch": 1.4331669439822519,
      "grad_norm": 0.04642242565751076,
      "learning_rate": 0.00017382716049382715,
      "loss": 0.5456,
      "step": 646
    },
    {
      "epoch": 1.435385468663339,
      "grad_norm": 0.03826535865664482,
      "learning_rate": 0.00017358024691358024,
      "loss": 0.5387,
      "step": 647
    },
    {
      "epoch": 1.437603993344426,
      "grad_norm": 0.04431036114692688,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.495,
      "step": 648
    },
    {
      "epoch": 1.439822518025513,
      "grad_norm": 0.0645177960395813,
      "learning_rate": 0.0001730864197530864,
      "loss": 0.5672,
      "step": 649
    },
    {
      "epoch": 1.4420410427066002,
      "grad_norm": 0.04858890175819397,
      "learning_rate": 0.0001728395061728395,
      "loss": 0.53,
      "step": 650
    },
    {
      "epoch": 1.4442595673876872,
      "grad_norm": 0.044151436537504196,
      "learning_rate": 0.00017259259259259257,
      "loss": 0.5933,
      "step": 651
    },
    {
      "epoch": 1.4464780920687743,
      "grad_norm": 0.06092476472258568,
      "learning_rate": 0.00017234567901234566,
      "loss": 0.6061,
      "step": 652
    },
    {
      "epoch": 1.4486966167498614,
      "grad_norm": 0.05232759192585945,
      "learning_rate": 0.00017209876543209873,
      "loss": 0.4751,
      "step": 653
    },
    {
      "epoch": 1.4509151414309485,
      "grad_norm": 0.05005029961466789,
      "learning_rate": 0.00017185185185185185,
      "loss": 0.5377,
      "step": 654
    },
    {
      "epoch": 1.4531336661120355,
      "grad_norm": 0.04101051390171051,
      "learning_rate": 0.00017160493827160492,
      "loss": 0.6025,
      "step": 655
    },
    {
      "epoch": 1.4553521907931226,
      "grad_norm": 0.060408495366573334,
      "learning_rate": 0.00017135802469135802,
      "loss": 0.5793,
      "step": 656
    },
    {
      "epoch": 1.4575707154742097,
      "grad_norm": 0.0368385910987854,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.5123,
      "step": 657
    },
    {
      "epoch": 1.4597892401552968,
      "grad_norm": 0.042350806295871735,
      "learning_rate": 0.00017086419753086418,
      "loss": 0.5873,
      "step": 658
    },
    {
      "epoch": 1.4620077648363838,
      "grad_norm": 0.04865170642733574,
      "learning_rate": 0.00017061728395061727,
      "loss": 0.7618,
      "step": 659
    },
    {
      "epoch": 1.464226289517471,
      "grad_norm": 0.04777724668383598,
      "learning_rate": 0.00017037037037037034,
      "loss": 0.5025,
      "step": 660
    },
    {
      "epoch": 1.466444814198558,
      "grad_norm": 0.0446874275803566,
      "learning_rate": 0.00017012345679012344,
      "loss": 0.7645,
      "step": 661
    },
    {
      "epoch": 1.468663338879645,
      "grad_norm": 0.04536648839712143,
      "learning_rate": 0.0001698765432098765,
      "loss": 0.6681,
      "step": 662
    },
    {
      "epoch": 1.4708818635607321,
      "grad_norm": 0.03377223387360573,
      "learning_rate": 0.00016962962962962963,
      "loss": 0.5645,
      "step": 663
    },
    {
      "epoch": 1.4731003882418192,
      "grad_norm": 0.047476280480623245,
      "learning_rate": 0.00016938271604938272,
      "loss": 0.6036,
      "step": 664
    },
    {
      "epoch": 1.4753189129229063,
      "grad_norm": 0.05199221894145012,
      "learning_rate": 0.0001691358024691358,
      "loss": 0.7161,
      "step": 665
    },
    {
      "epoch": 1.4775374376039934,
      "grad_norm": 0.06521202623844147,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.7171,
      "step": 666
    },
    {
      "epoch": 1.4797559622850804,
      "grad_norm": 0.03884696960449219,
      "learning_rate": 0.00016864197530864195,
      "loss": 0.577,
      "step": 667
    },
    {
      "epoch": 1.4819744869661675,
      "grad_norm": 0.06097685545682907,
      "learning_rate": 0.00016839506172839505,
      "loss": 0.4578,
      "step": 668
    },
    {
      "epoch": 1.4841930116472546,
      "grad_norm": 0.03534045070409775,
      "learning_rate": 0.00016814814814814812,
      "loss": 0.6317,
      "step": 669
    },
    {
      "epoch": 1.4864115363283417,
      "grad_norm": 0.05629810690879822,
      "learning_rate": 0.0001679012345679012,
      "loss": 0.5951,
      "step": 670
    },
    {
      "epoch": 1.4886300610094287,
      "grad_norm": 0.030927609652280807,
      "learning_rate": 0.00016765432098765433,
      "loss": 0.5492,
      "step": 671
    },
    {
      "epoch": 1.4908485856905158,
      "grad_norm": 0.03698636218905449,
      "learning_rate": 0.0001674074074074074,
      "loss": 0.5787,
      "step": 672
    },
    {
      "epoch": 1.493067110371603,
      "grad_norm": 0.03189109265804291,
      "learning_rate": 0.0001671604938271605,
      "loss": 0.5223,
      "step": 673
    },
    {
      "epoch": 1.49528563505269,
      "grad_norm": 0.044681843370199203,
      "learning_rate": 0.00016691358024691357,
      "loss": 0.5461,
      "step": 674
    },
    {
      "epoch": 1.497504159733777,
      "grad_norm": 0.03493814915418625,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.4822,
      "step": 675
    },
    {
      "epoch": 1.4997226844148641,
      "grad_norm": 0.06603062897920609,
      "learning_rate": 0.00016641975308641973,
      "loss": 0.5415,
      "step": 676
    },
    {
      "epoch": 1.5019412090959512,
      "grad_norm": 0.05028144642710686,
      "learning_rate": 0.00016617283950617282,
      "loss": 0.7155,
      "step": 677
    },
    {
      "epoch": 1.5041597337770383,
      "grad_norm": 0.07729141414165497,
      "learning_rate": 0.0001659259259259259,
      "loss": 0.4679,
      "step": 678
    },
    {
      "epoch": 1.5063782584581253,
      "grad_norm": 0.039953626692295074,
      "learning_rate": 0.000165679012345679,
      "loss": 0.5613,
      "step": 679
    },
    {
      "epoch": 1.5085967831392124,
      "grad_norm": 0.04812348634004593,
      "learning_rate": 0.0001654320987654321,
      "loss": 0.5224,
      "step": 680
    },
    {
      "epoch": 1.5108153078202995,
      "grad_norm": 0.09428009390830994,
      "learning_rate": 0.00016518518518518518,
      "loss": 0.6451,
      "step": 681
    },
    {
      "epoch": 1.5130338325013866,
      "grad_norm": 0.0929945856332779,
      "learning_rate": 0.00016493827160493827,
      "loss": 0.8078,
      "step": 682
    },
    {
      "epoch": 1.5152523571824736,
      "grad_norm": 0.04322494566440582,
      "learning_rate": 0.00016469135802469134,
      "loss": 0.5784,
      "step": 683
    },
    {
      "epoch": 1.5174708818635607,
      "grad_norm": 0.05272587388753891,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.6097,
      "step": 684
    },
    {
      "epoch": 1.5196894065446478,
      "grad_norm": 0.0571681410074234,
      "learning_rate": 0.0001641975308641975,
      "loss": 0.6385,
      "step": 685
    },
    {
      "epoch": 1.5219079312257349,
      "grad_norm": 0.051645535975694656,
      "learning_rate": 0.0001639506172839506,
      "loss": 0.5052,
      "step": 686
    },
    {
      "epoch": 1.524126455906822,
      "grad_norm": 0.09513190388679504,
      "learning_rate": 0.0001637037037037037,
      "loss": 0.655,
      "step": 687
    },
    {
      "epoch": 1.526344980587909,
      "grad_norm": 0.03873327374458313,
      "learning_rate": 0.00016345679012345676,
      "loss": 0.7225,
      "step": 688
    },
    {
      "epoch": 1.528563505268996,
      "grad_norm": 0.04208820313215256,
      "learning_rate": 0.00016320987654320988,
      "loss": 0.5114,
      "step": 689
    },
    {
      "epoch": 1.5307820299500832,
      "grad_norm": 0.0393582321703434,
      "learning_rate": 0.00016296296296296295,
      "loss": 0.5352,
      "step": 690
    },
    {
      "epoch": 1.5330005546311702,
      "grad_norm": 0.0460151843726635,
      "learning_rate": 0.00016271604938271605,
      "loss": 0.553,
      "step": 691
    },
    {
      "epoch": 1.5352190793122573,
      "grad_norm": 0.07763145118951797,
      "learning_rate": 0.00016246913580246912,
      "loss": 0.6321,
      "step": 692
    },
    {
      "epoch": 1.5374376039933444,
      "grad_norm": 0.04139331728219986,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.6486,
      "step": 693
    },
    {
      "epoch": 1.5396561286744315,
      "grad_norm": 0.05501311644911766,
      "learning_rate": 0.0001619753086419753,
      "loss": 0.6876,
      "step": 694
    },
    {
      "epoch": 1.5418746533555185,
      "grad_norm": 0.0505472868680954,
      "learning_rate": 0.00016172839506172837,
      "loss": 0.6075,
      "step": 695
    },
    {
      "epoch": 1.5440931780366056,
      "grad_norm": 0.04209991171956062,
      "learning_rate": 0.00016148148148148147,
      "loss": 0.6525,
      "step": 696
    },
    {
      "epoch": 1.5463117027176927,
      "grad_norm": 0.04096788167953491,
      "learning_rate": 0.00016123456790123454,
      "loss": 0.5865,
      "step": 697
    },
    {
      "epoch": 1.5485302273987798,
      "grad_norm": 0.042752716690301895,
      "learning_rate": 0.00016098765432098766,
      "loss": 0.5326,
      "step": 698
    },
    {
      "epoch": 1.5507487520798668,
      "grad_norm": 0.04859471693634987,
      "learning_rate": 0.00016074074074074073,
      "loss": 0.679,
      "step": 699
    },
    {
      "epoch": 1.552967276760954,
      "grad_norm": 0.05045028403401375,
      "learning_rate": 0.00016049382716049382,
      "loss": 0.5411,
      "step": 700
    },
    {
      "epoch": 1.555185801442041,
      "grad_norm": 0.053076088428497314,
      "learning_rate": 0.00016024691358024692,
      "loss": 0.6174,
      "step": 701
    },
    {
      "epoch": 1.557404326123128,
      "grad_norm": 0.06258540600538254,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.6757,
      "step": 702
    },
    {
      "epoch": 1.5596228508042151,
      "grad_norm": 0.07342540472745895,
      "learning_rate": 0.00015975308641975308,
      "loss": 0.8069,
      "step": 703
    },
    {
      "epoch": 1.5618413754853022,
      "grad_norm": 0.05042373761534691,
      "learning_rate": 0.00015950617283950615,
      "loss": 0.6243,
      "step": 704
    },
    {
      "epoch": 1.5640599001663893,
      "grad_norm": 0.06581959128379822,
      "learning_rate": 0.00015925925925925924,
      "loss": 0.6881,
      "step": 705
    },
    {
      "epoch": 1.5662784248474764,
      "grad_norm": 0.0643521100282669,
      "learning_rate": 0.0001590123456790123,
      "loss": 0.559,
      "step": 706
    },
    {
      "epoch": 1.5684969495285634,
      "grad_norm": 0.04091258347034454,
      "learning_rate": 0.00015876543209876543,
      "loss": 0.5252,
      "step": 707
    },
    {
      "epoch": 1.5707154742096505,
      "grad_norm": 0.038373272866010666,
      "learning_rate": 0.0001585185185185185,
      "loss": 0.4588,
      "step": 708
    },
    {
      "epoch": 1.5729339988907376,
      "grad_norm": 0.03779299929738045,
      "learning_rate": 0.0001582716049382716,
      "loss": 0.5913,
      "step": 709
    },
    {
      "epoch": 1.5751525235718247,
      "grad_norm": 0.03291252255439758,
      "learning_rate": 0.0001580246913580247,
      "loss": 0.4725,
      "step": 710
    },
    {
      "epoch": 1.5773710482529117,
      "grad_norm": 0.06149889901280403,
      "learning_rate": 0.00015777777777777776,
      "loss": 0.7636,
      "step": 711
    },
    {
      "epoch": 1.5795895729339988,
      "grad_norm": 0.03663446381688118,
      "learning_rate": 0.00015753086419753086,
      "loss": 0.5255,
      "step": 712
    },
    {
      "epoch": 1.581808097615086,
      "grad_norm": 0.04454001411795616,
      "learning_rate": 0.00015728395061728392,
      "loss": 0.4245,
      "step": 713
    },
    {
      "epoch": 1.584026622296173,
      "grad_norm": 0.04806433990597725,
      "learning_rate": 0.00015703703703703702,
      "loss": 0.6949,
      "step": 714
    },
    {
      "epoch": 1.58624514697726,
      "grad_norm": 0.03919399157166481,
      "learning_rate": 0.0001567901234567901,
      "loss": 0.4817,
      "step": 715
    },
    {
      "epoch": 1.5884636716583471,
      "grad_norm": 0.0396035872399807,
      "learning_rate": 0.0001565432098765432,
      "loss": 0.6889,
      "step": 716
    },
    {
      "epoch": 1.5906821963394342,
      "grad_norm": 0.03566046431660652,
      "learning_rate": 0.0001562962962962963,
      "loss": 0.4745,
      "step": 717
    },
    {
      "epoch": 1.5929007210205213,
      "grad_norm": 0.050695475190877914,
      "learning_rate": 0.00015604938271604937,
      "loss": 0.5907,
      "step": 718
    },
    {
      "epoch": 1.5951192457016083,
      "grad_norm": 0.03896141052246094,
      "learning_rate": 0.00015580246913580247,
      "loss": 0.577,
      "step": 719
    },
    {
      "epoch": 1.5973377703826954,
      "grad_norm": 0.048122502863407135,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.6956,
      "step": 720
    },
    {
      "epoch": 1.5995562950637825,
      "grad_norm": 0.03698781877756119,
      "learning_rate": 0.00015530864197530863,
      "loss": 0.6193,
      "step": 721
    },
    {
      "epoch": 1.6017748197448696,
      "grad_norm": 0.04659617319703102,
      "learning_rate": 0.0001550617283950617,
      "loss": 0.59,
      "step": 722
    },
    {
      "epoch": 1.6039933444259566,
      "grad_norm": 0.060981158167123795,
      "learning_rate": 0.0001548148148148148,
      "loss": 0.4466,
      "step": 723
    },
    {
      "epoch": 1.6062118691070437,
      "grad_norm": 0.03238467872142792,
      "learning_rate": 0.00015456790123456792,
      "loss": 0.7607,
      "step": 724
    },
    {
      "epoch": 1.6084303937881308,
      "grad_norm": 0.04374478757381439,
      "learning_rate": 0.00015432098765432098,
      "loss": 0.5256,
      "step": 725
    },
    {
      "epoch": 1.6106489184692179,
      "grad_norm": 0.04634390398859978,
      "learning_rate": 0.00015407407407407408,
      "loss": 0.6286,
      "step": 726
    },
    {
      "epoch": 1.612867443150305,
      "grad_norm": 0.04896163195371628,
      "learning_rate": 0.00015382716049382715,
      "loss": 0.6148,
      "step": 727
    },
    {
      "epoch": 1.615085967831392,
      "grad_norm": 0.04786650836467743,
      "learning_rate": 0.00015358024691358024,
      "loss": 0.7101,
      "step": 728
    },
    {
      "epoch": 1.617304492512479,
      "grad_norm": 0.04435646906495094,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.6338,
      "step": 729
    },
    {
      "epoch": 1.6195230171935662,
      "grad_norm": 0.044978030025959015,
      "learning_rate": 0.0001530864197530864,
      "loss": 0.6362,
      "step": 730
    },
    {
      "epoch": 1.6217415418746532,
      "grad_norm": 0.037433214485645294,
      "learning_rate": 0.00015283950617283947,
      "loss": 0.5452,
      "step": 731
    },
    {
      "epoch": 1.6239600665557403,
      "grad_norm": 0.049420710653066635,
      "learning_rate": 0.00015259259259259257,
      "loss": 0.713,
      "step": 732
    },
    {
      "epoch": 1.6261785912368274,
      "grad_norm": 0.033680714666843414,
      "learning_rate": 0.0001523456790123457,
      "loss": 0.4439,
      "step": 733
    },
    {
      "epoch": 1.6283971159179145,
      "grad_norm": 0.04289427399635315,
      "learning_rate": 0.00015209876543209873,
      "loss": 0.5518,
      "step": 734
    },
    {
      "epoch": 1.6306156405990015,
      "grad_norm": 0.04142884537577629,
      "learning_rate": 0.00015185185185185185,
      "loss": 0.5738,
      "step": 735
    },
    {
      "epoch": 1.6328341652800886,
      "grad_norm": 0.036771342158317566,
      "learning_rate": 0.00015160493827160492,
      "loss": 0.5506,
      "step": 736
    },
    {
      "epoch": 1.6350526899611757,
      "grad_norm": 0.03521214425563812,
      "learning_rate": 0.00015135802469135802,
      "loss": 0.5526,
      "step": 737
    },
    {
      "epoch": 1.6372712146422628,
      "grad_norm": 0.039879295974969864,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.5416,
      "step": 738
    },
    {
      "epoch": 1.6394897393233498,
      "grad_norm": 0.03407967835664749,
      "learning_rate": 0.00015086419753086418,
      "loss": 0.5705,
      "step": 739
    },
    {
      "epoch": 1.641708264004437,
      "grad_norm": 0.05392766743898392,
      "learning_rate": 0.00015061728395061728,
      "loss": 0.638,
      "step": 740
    },
    {
      "epoch": 1.643926788685524,
      "grad_norm": 0.045433711260557175,
      "learning_rate": 0.00015037037037037034,
      "loss": 0.6796,
      "step": 741
    },
    {
      "epoch": 1.646145313366611,
      "grad_norm": 0.041845738887786865,
      "learning_rate": 0.00015012345679012347,
      "loss": 0.4509,
      "step": 742
    },
    {
      "epoch": 1.6483638380476981,
      "grad_norm": 0.041106581687927246,
      "learning_rate": 0.00014987654320987653,
      "loss": 0.6648,
      "step": 743
    },
    {
      "epoch": 1.6505823627287852,
      "grad_norm": 0.044286709278821945,
      "learning_rate": 0.00014962962962962963,
      "loss": 0.6102,
      "step": 744
    },
    {
      "epoch": 1.6528008874098723,
      "grad_norm": 0.055176541209220886,
      "learning_rate": 0.0001493827160493827,
      "loss": 0.596,
      "step": 745
    },
    {
      "epoch": 1.6550194120909594,
      "grad_norm": 0.03873584046959877,
      "learning_rate": 0.0001491358024691358,
      "loss": 0.6212,
      "step": 746
    },
    {
      "epoch": 1.6572379367720464,
      "grad_norm": 0.04418507218360901,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.6547,
      "step": 747
    },
    {
      "epoch": 1.6594564614531335,
      "grad_norm": 0.03445843607187271,
      "learning_rate": 0.00014864197530864196,
      "loss": 0.5448,
      "step": 748
    },
    {
      "epoch": 1.6616749861342206,
      "grad_norm": 0.03362668305635452,
      "learning_rate": 0.00014839506172839505,
      "loss": 0.5085,
      "step": 749
    },
    {
      "epoch": 1.6638935108153077,
      "grad_norm": 0.031335242092609406,
      "learning_rate": 0.00014814814814814812,
      "loss": 0.3908,
      "step": 750
    },
    {
      "epoch": 1.6661120354963947,
      "grad_norm": 0.035022418946027756,
      "learning_rate": 0.00014790123456790124,
      "loss": 0.4678,
      "step": 751
    },
    {
      "epoch": 1.6683305601774818,
      "grad_norm": 0.04569832980632782,
      "learning_rate": 0.0001476543209876543,
      "loss": 0.5006,
      "step": 752
    },
    {
      "epoch": 1.670549084858569,
      "grad_norm": 0.045884065330028534,
      "learning_rate": 0.0001474074074074074,
      "loss": 0.5539,
      "step": 753
    },
    {
      "epoch": 1.672767609539656,
      "grad_norm": 0.04756151884794235,
      "learning_rate": 0.00014716049382716047,
      "loss": 0.6092,
      "step": 754
    },
    {
      "epoch": 1.674986134220743,
      "grad_norm": 0.034285999834537506,
      "learning_rate": 0.00014691358024691357,
      "loss": 0.6215,
      "step": 755
    },
    {
      "epoch": 1.6772046589018301,
      "grad_norm": 0.03463876619935036,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.6968,
      "step": 756
    },
    {
      "epoch": 1.6794231835829172,
      "grad_norm": 0.034965332597494125,
      "learning_rate": 0.00014641975308641973,
      "loss": 0.6039,
      "step": 757
    },
    {
      "epoch": 1.6816417082640043,
      "grad_norm": 0.03379840403795242,
      "learning_rate": 0.00014617283950617283,
      "loss": 0.5809,
      "step": 758
    },
    {
      "epoch": 1.6838602329450914,
      "grad_norm": 0.0410749614238739,
      "learning_rate": 0.00014592592592592592,
      "loss": 0.5944,
      "step": 759
    },
    {
      "epoch": 1.6860787576261784,
      "grad_norm": 0.04708585515618324,
      "learning_rate": 0.000145679012345679,
      "loss": 0.6248,
      "step": 760
    },
    {
      "epoch": 1.6882972823072657,
      "grad_norm": 0.04456539452075958,
      "learning_rate": 0.00014543209876543208,
      "loss": 0.5921,
      "step": 761
    },
    {
      "epoch": 1.6905158069883528,
      "grad_norm": 0.03964368253946304,
      "learning_rate": 0.00014518518518518518,
      "loss": 0.5753,
      "step": 762
    },
    {
      "epoch": 1.6927343316694399,
      "grad_norm": 0.06042619049549103,
      "learning_rate": 0.00014493827160493825,
      "loss": 0.9527,
      "step": 763
    },
    {
      "epoch": 1.694952856350527,
      "grad_norm": 0.061779800802469254,
      "learning_rate": 0.00014469135802469134,
      "loss": 0.6003,
      "step": 764
    },
    {
      "epoch": 1.697171381031614,
      "grad_norm": 0.13690075278282166,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.7283,
      "step": 765
    },
    {
      "epoch": 1.699389905712701,
      "grad_norm": 0.040135130286216736,
      "learning_rate": 0.00014419753086419753,
      "loss": 0.5452,
      "step": 766
    },
    {
      "epoch": 1.7016084303937882,
      "grad_norm": 0.04853465408086777,
      "learning_rate": 0.0001439506172839506,
      "loss": 0.822,
      "step": 767
    },
    {
      "epoch": 1.7038269550748752,
      "grad_norm": 0.032917335629463196,
      "learning_rate": 0.0001437037037037037,
      "loss": 0.4619,
      "step": 768
    },
    {
      "epoch": 1.7060454797559623,
      "grad_norm": 0.08757856488227844,
      "learning_rate": 0.00014345679012345676,
      "loss": 0.5685,
      "step": 769
    },
    {
      "epoch": 1.7082640044370494,
      "grad_norm": 0.05815499648451805,
      "learning_rate": 0.00014320987654320986,
      "loss": 0.6849,
      "step": 770
    },
    {
      "epoch": 1.7104825291181365,
      "grad_norm": 0.03439812362194061,
      "learning_rate": 0.00014296296296296295,
      "loss": 0.5427,
      "step": 771
    },
    {
      "epoch": 1.7127010537992235,
      "grad_norm": 0.05457419902086258,
      "learning_rate": 0.00014271604938271602,
      "loss": 0.5719,
      "step": 772
    },
    {
      "epoch": 1.7149195784803106,
      "grad_norm": 0.05357319489121437,
      "learning_rate": 0.00014246913580246912,
      "loss": 0.6741,
      "step": 773
    },
    {
      "epoch": 1.7171381031613977,
      "grad_norm": 0.04329128935933113,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.6617,
      "step": 774
    },
    {
      "epoch": 1.7193566278424848,
      "grad_norm": 0.04493206739425659,
      "learning_rate": 0.0001419753086419753,
      "loss": 0.4818,
      "step": 775
    },
    {
      "epoch": 1.7215751525235718,
      "grad_norm": 0.046465810388326645,
      "learning_rate": 0.00014172839506172838,
      "loss": 0.6314,
      "step": 776
    },
    {
      "epoch": 1.723793677204659,
      "grad_norm": 0.04128435626626015,
      "learning_rate": 0.00014148148148148147,
      "loss": 0.7185,
      "step": 777
    },
    {
      "epoch": 1.726012201885746,
      "grad_norm": 0.04832012206315994,
      "learning_rate": 0.00014123456790123454,
      "loss": 0.6836,
      "step": 778
    },
    {
      "epoch": 1.728230726566833,
      "grad_norm": 0.04681311175227165,
      "learning_rate": 0.00014098765432098763,
      "loss": 0.5593,
      "step": 779
    },
    {
      "epoch": 1.7304492512479202,
      "grad_norm": 0.04425084590911865,
      "learning_rate": 0.00014074074074074073,
      "loss": 0.619,
      "step": 780
    },
    {
      "epoch": 1.7326677759290072,
      "grad_norm": 0.03669988736510277,
      "learning_rate": 0.00014049382716049382,
      "loss": 0.5538,
      "step": 781
    },
    {
      "epoch": 1.7348863006100943,
      "grad_norm": 0.06397844105958939,
      "learning_rate": 0.0001402469135802469,
      "loss": 0.5901,
      "step": 782
    },
    {
      "epoch": 1.7371048252911814,
      "grad_norm": 0.05260743200778961,
      "learning_rate": 0.00014,
      "loss": 0.6042,
      "step": 783
    },
    {
      "epoch": 1.7393233499722685,
      "grad_norm": 0.0451732873916626,
      "learning_rate": 0.00013975308641975308,
      "loss": 0.5717,
      "step": 784
    },
    {
      "epoch": 1.7415418746533555,
      "grad_norm": 0.05897792428731918,
      "learning_rate": 0.00013950617283950615,
      "loss": 0.6149,
      "step": 785
    },
    {
      "epoch": 1.7437603993344426,
      "grad_norm": 0.037755634635686874,
      "learning_rate": 0.00013925925925925925,
      "loss": 0.628,
      "step": 786
    },
    {
      "epoch": 1.7459789240155297,
      "grad_norm": 0.03767247870564461,
      "learning_rate": 0.00013901234567901231,
      "loss": 0.4848,
      "step": 787
    },
    {
      "epoch": 1.7481974486966168,
      "grad_norm": 0.05688023567199707,
      "learning_rate": 0.00013876543209876544,
      "loss": 0.7702,
      "step": 788
    },
    {
      "epoch": 1.7504159733777038,
      "grad_norm": 0.08141244947910309,
      "learning_rate": 0.0001385185185185185,
      "loss": 0.563,
      "step": 789
    },
    {
      "epoch": 1.752634498058791,
      "grad_norm": 0.03885933384299278,
      "learning_rate": 0.0001382716049382716,
      "loss": 0.5063,
      "step": 790
    },
    {
      "epoch": 1.754853022739878,
      "grad_norm": 0.03937618061900139,
      "learning_rate": 0.00013802469135802467,
      "loss": 0.6058,
      "step": 791
    },
    {
      "epoch": 1.757071547420965,
      "grad_norm": 0.04535830393433571,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.4731,
      "step": 792
    },
    {
      "epoch": 1.7592900721020521,
      "grad_norm": 0.05277817323803902,
      "learning_rate": 0.00013753086419753086,
      "loss": 0.4983,
      "step": 793
    },
    {
      "epoch": 1.7615085967831392,
      "grad_norm": 0.041434381157159805,
      "learning_rate": 0.00013728395061728393,
      "loss": 0.606,
      "step": 794
    },
    {
      "epoch": 1.7637271214642263,
      "grad_norm": 0.046086184680461884,
      "learning_rate": 0.00013703703703703702,
      "loss": 0.6235,
      "step": 795
    },
    {
      "epoch": 1.7659456461453134,
      "grad_norm": 0.04551013559103012,
      "learning_rate": 0.00013679012345679012,
      "loss": 0.6113,
      "step": 796
    },
    {
      "epoch": 1.7681641708264004,
      "grad_norm": 0.038690824061632156,
      "learning_rate": 0.0001365432098765432,
      "loss": 0.5983,
      "step": 797
    },
    {
      "epoch": 1.7703826955074875,
      "grad_norm": 0.045511644333601,
      "learning_rate": 0.00013629629629629628,
      "loss": 0.5647,
      "step": 798
    },
    {
      "epoch": 1.7726012201885746,
      "grad_norm": 0.042262256145477295,
      "learning_rate": 0.00013604938271604937,
      "loss": 0.6379,
      "step": 799
    },
    {
      "epoch": 1.7748197448696617,
      "grad_norm": 0.05710117146372795,
      "learning_rate": 0.00013580246913580244,
      "loss": 0.5428,
      "step": 800
    },
    {
      "epoch": 1.7770382695507487,
      "grad_norm": 0.058080948889255524,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.5121,
      "step": 801
    },
    {
      "epoch": 1.7792567942318358,
      "grad_norm": 0.051045604050159454,
      "learning_rate": 0.00013530864197530863,
      "loss": 0.525,
      "step": 802
    },
    {
      "epoch": 1.7814753189129229,
      "grad_norm": 0.03683939576148987,
      "learning_rate": 0.00013506172839506173,
      "loss": 0.4955,
      "step": 803
    },
    {
      "epoch": 1.78369384359401,
      "grad_norm": 0.0535573735833168,
      "learning_rate": 0.0001348148148148148,
      "loss": 0.6173,
      "step": 804
    },
    {
      "epoch": 1.785912368275097,
      "grad_norm": 0.04746395722031593,
      "learning_rate": 0.0001345679012345679,
      "loss": 0.4894,
      "step": 805
    },
    {
      "epoch": 1.788130892956184,
      "grad_norm": 0.03501688316464424,
      "learning_rate": 0.00013432098765432099,
      "loss": 0.5487,
      "step": 806
    },
    {
      "epoch": 1.7903494176372712,
      "grad_norm": 0.06083143875002861,
      "learning_rate": 0.00013407407407407405,
      "loss": 0.6118,
      "step": 807
    },
    {
      "epoch": 1.7925679423183583,
      "grad_norm": 0.05251947417855263,
      "learning_rate": 0.00013382716049382715,
      "loss": 0.6776,
      "step": 808
    },
    {
      "epoch": 1.7947864669994453,
      "grad_norm": 0.03290028125047684,
      "learning_rate": 0.00013358024691358022,
      "loss": 0.495,
      "step": 809
    },
    {
      "epoch": 1.7970049916805324,
      "grad_norm": 0.040215376764535904,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.5454,
      "step": 810
    },
    {
      "epoch": 1.7992235163616195,
      "grad_norm": 0.036370810121297836,
      "learning_rate": 0.0001330864197530864,
      "loss": 0.5118,
      "step": 811
    },
    {
      "epoch": 1.8014420410427066,
      "grad_norm": 0.03826798498630524,
      "learning_rate": 0.0001328395061728395,
      "loss": 0.6034,
      "step": 812
    },
    {
      "epoch": 1.8036605657237936,
      "grad_norm": 0.041245292872190475,
      "learning_rate": 0.00013259259259259257,
      "loss": 0.6121,
      "step": 813
    },
    {
      "epoch": 1.8058790904048807,
      "grad_norm": 0.046567823737859726,
      "learning_rate": 0.00013234567901234567,
      "loss": 0.5875,
      "step": 814
    },
    {
      "epoch": 1.8080976150859678,
      "grad_norm": 0.05198601260781288,
      "learning_rate": 0.00013209876543209876,
      "loss": 0.54,
      "step": 815
    },
    {
      "epoch": 1.8103161397670549,
      "grad_norm": 0.061592765152454376,
      "learning_rate": 0.00013185185185185183,
      "loss": 0.7984,
      "step": 816
    },
    {
      "epoch": 1.8125346644481422,
      "grad_norm": 0.050767313688993454,
      "learning_rate": 0.00013160493827160492,
      "loss": 0.5678,
      "step": 817
    },
    {
      "epoch": 1.8147531891292292,
      "grad_norm": 0.049739204347133636,
      "learning_rate": 0.00013135802469135802,
      "loss": 0.6165,
      "step": 818
    },
    {
      "epoch": 1.8169717138103163,
      "grad_norm": 0.04584597423672676,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.5989,
      "step": 819
    },
    {
      "epoch": 1.8191902384914034,
      "grad_norm": 0.0730195865035057,
      "learning_rate": 0.00013086419753086418,
      "loss": 0.5609,
      "step": 820
    },
    {
      "epoch": 1.8214087631724905,
      "grad_norm": 0.039064712822437286,
      "learning_rate": 0.00013061728395061728,
      "loss": 0.6131,
      "step": 821
    },
    {
      "epoch": 1.8236272878535775,
      "grad_norm": 0.06072591617703438,
      "learning_rate": 0.00013037037037037035,
      "loss": 0.6418,
      "step": 822
    },
    {
      "epoch": 1.8258458125346646,
      "grad_norm": 0.037326645106077194,
      "learning_rate": 0.00013012345679012344,
      "loss": 0.5248,
      "step": 823
    },
    {
      "epoch": 1.8280643372157517,
      "grad_norm": 0.05533823370933533,
      "learning_rate": 0.00012987654320987654,
      "loss": 0.7194,
      "step": 824
    },
    {
      "epoch": 1.8302828618968388,
      "grad_norm": 0.05212606489658356,
      "learning_rate": 0.0001296296296296296,
      "loss": 0.6055,
      "step": 825
    },
    {
      "epoch": 1.8325013865779258,
      "grad_norm": 0.06374059617519379,
      "learning_rate": 0.0001293827160493827,
      "loss": 0.6653,
      "step": 826
    },
    {
      "epoch": 1.834719911259013,
      "grad_norm": 0.04174331575632095,
      "learning_rate": 0.0001291358024691358,
      "loss": 0.6794,
      "step": 827
    },
    {
      "epoch": 1.8369384359401,
      "grad_norm": 0.041831228882074356,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.5143,
      "step": 828
    },
    {
      "epoch": 1.839156960621187,
      "grad_norm": 0.04617135971784592,
      "learning_rate": 0.00012864197530864196,
      "loss": 0.6311,
      "step": 829
    },
    {
      "epoch": 1.8413754853022741,
      "grad_norm": 0.061536844819784164,
      "learning_rate": 0.00012839506172839505,
      "loss": 0.6367,
      "step": 830
    },
    {
      "epoch": 1.8435940099833612,
      "grad_norm": 0.08150152117013931,
      "learning_rate": 0.00012814814814814812,
      "loss": 0.6455,
      "step": 831
    },
    {
      "epoch": 1.8458125346644483,
      "grad_norm": 0.04443927854299545,
      "learning_rate": 0.00012790123456790122,
      "loss": 0.6065,
      "step": 832
    },
    {
      "epoch": 1.8480310593455354,
      "grad_norm": 0.07198163121938705,
      "learning_rate": 0.0001276543209876543,
      "loss": 0.7068,
      "step": 833
    },
    {
      "epoch": 1.8502495840266224,
      "grad_norm": 0.0361124686896801,
      "learning_rate": 0.0001274074074074074,
      "loss": 0.4917,
      "step": 834
    },
    {
      "epoch": 1.8524681087077095,
      "grad_norm": 0.04808807000517845,
      "learning_rate": 0.00012716049382716047,
      "loss": 0.5813,
      "step": 835
    },
    {
      "epoch": 1.8546866333887966,
      "grad_norm": 0.044794414192438126,
      "learning_rate": 0.00012691358024691357,
      "loss": 0.6987,
      "step": 836
    },
    {
      "epoch": 1.8569051580698837,
      "grad_norm": 0.05478532239794731,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.6436,
      "step": 837
    },
    {
      "epoch": 1.8591236827509707,
      "grad_norm": 0.060514405369758606,
      "learning_rate": 0.00012641975308641973,
      "loss": 0.5956,
      "step": 838
    },
    {
      "epoch": 1.8613422074320578,
      "grad_norm": 0.035892702639102936,
      "learning_rate": 0.00012617283950617283,
      "loss": 0.481,
      "step": 839
    },
    {
      "epoch": 1.8635607321131449,
      "grad_norm": 0.03388211131095886,
      "learning_rate": 0.0001259259259259259,
      "loss": 0.5782,
      "step": 840
    },
    {
      "epoch": 1.865779256794232,
      "grad_norm": 0.05917976796627045,
      "learning_rate": 0.00012567901234567902,
      "loss": 0.7009,
      "step": 841
    },
    {
      "epoch": 1.867997781475319,
      "grad_norm": 0.04124234616756439,
      "learning_rate": 0.00012543209876543209,
      "loss": 0.6677,
      "step": 842
    },
    {
      "epoch": 1.870216306156406,
      "grad_norm": 0.09152578562498093,
      "learning_rate": 0.00012518518518518518,
      "loss": 0.5645,
      "step": 843
    },
    {
      "epoch": 1.8724348308374932,
      "grad_norm": 0.044471122324466705,
      "learning_rate": 0.00012493827160493825,
      "loss": 0.731,
      "step": 844
    },
    {
      "epoch": 1.8746533555185803,
      "grad_norm": 0.03328584507107735,
      "learning_rate": 0.00012469135802469134,
      "loss": 0.5237,
      "step": 845
    },
    {
      "epoch": 1.8768718801996673,
      "grad_norm": 0.04293752834200859,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.6851,
      "step": 846
    },
    {
      "epoch": 1.8790904048807544,
      "grad_norm": 0.03816654160618782,
      "learning_rate": 0.0001241975308641975,
      "loss": 0.5566,
      "step": 847
    },
    {
      "epoch": 1.8813089295618415,
      "grad_norm": 0.04423398897051811,
      "learning_rate": 0.0001239506172839506,
      "loss": 0.6169,
      "step": 848
    },
    {
      "epoch": 1.8835274542429286,
      "grad_norm": 0.04283075034618378,
      "learning_rate": 0.0001237037037037037,
      "loss": 0.571,
      "step": 849
    },
    {
      "epoch": 1.8857459789240156,
      "grad_norm": 0.04268324747681618,
      "learning_rate": 0.0001234567901234568,
      "loss": 0.5879,
      "step": 850
    },
    {
      "epoch": 1.8879645036051027,
      "grad_norm": 0.03956582769751549,
      "learning_rate": 0.00012320987654320986,
      "loss": 0.6111,
      "step": 851
    },
    {
      "epoch": 1.8901830282861898,
      "grad_norm": 0.03609675541520119,
      "learning_rate": 0.00012296296296296296,
      "loss": 0.5647,
      "step": 852
    },
    {
      "epoch": 1.8924015529672769,
      "grad_norm": 0.04365142434835434,
      "learning_rate": 0.00012271604938271602,
      "loss": 0.4536,
      "step": 853
    },
    {
      "epoch": 1.894620077648364,
      "grad_norm": 0.044804953038692474,
      "learning_rate": 0.00012246913580246912,
      "loss": 0.5501,
      "step": 854
    },
    {
      "epoch": 1.896838602329451,
      "grad_norm": 0.03310094401240349,
      "learning_rate": 0.00012222222222222221,
      "loss": 0.7541,
      "step": 855
    },
    {
      "epoch": 1.899057127010538,
      "grad_norm": 0.031052658334374428,
      "learning_rate": 0.00012197530864197531,
      "loss": 0.4813,
      "step": 856
    },
    {
      "epoch": 1.9012756516916252,
      "grad_norm": 0.05651732161641121,
      "learning_rate": 0.00012172839506172839,
      "loss": 0.4315,
      "step": 857
    },
    {
      "epoch": 1.9034941763727122,
      "grad_norm": 0.05741611868143082,
      "learning_rate": 0.00012148148148148147,
      "loss": 0.7092,
      "step": 858
    },
    {
      "epoch": 1.9057127010537993,
      "grad_norm": 0.05942470580339432,
      "learning_rate": 0.00012123456790123455,
      "loss": 0.7611,
      "step": 859
    },
    {
      "epoch": 1.9079312257348864,
      "grad_norm": 0.04893913492560387,
      "learning_rate": 0.00012098765432098764,
      "loss": 0.614,
      "step": 860
    },
    {
      "epoch": 1.9101497504159735,
      "grad_norm": 0.02902265265583992,
      "learning_rate": 0.00012074074074074073,
      "loss": 0.792,
      "step": 861
    },
    {
      "epoch": 1.9123682750970605,
      "grad_norm": 0.03685220330953598,
      "learning_rate": 0.00012049382716049381,
      "loss": 0.4937,
      "step": 862
    },
    {
      "epoch": 1.9145867997781476,
      "grad_norm": 0.04143933951854706,
      "learning_rate": 0.0001202469135802469,
      "loss": 0.4741,
      "step": 863
    },
    {
      "epoch": 1.9168053244592347,
      "grad_norm": 0.060860488563776016,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.5626,
      "step": 864
    },
    {
      "epoch": 1.9190238491403218,
      "grad_norm": 0.032612916082143784,
      "learning_rate": 0.00011975308641975308,
      "loss": 0.5621,
      "step": 865
    },
    {
      "epoch": 1.9212423738214088,
      "grad_norm": 0.05585114657878876,
      "learning_rate": 0.00011950617283950617,
      "loss": 0.5969,
      "step": 866
    },
    {
      "epoch": 1.923460898502496,
      "grad_norm": 0.05776425823569298,
      "learning_rate": 0.00011925925925925925,
      "loss": 0.5067,
      "step": 867
    },
    {
      "epoch": 1.925679423183583,
      "grad_norm": 0.05040394142270088,
      "learning_rate": 0.00011901234567901233,
      "loss": 0.7389,
      "step": 868
    },
    {
      "epoch": 1.92789794786467,
      "grad_norm": 0.056915003806352615,
      "learning_rate": 0.00011876543209876541,
      "loss": 0.6051,
      "step": 869
    },
    {
      "epoch": 1.9301164725457571,
      "grad_norm": 0.05292448773980141,
      "learning_rate": 0.0001185185185185185,
      "loss": 0.7834,
      "step": 870
    },
    {
      "epoch": 1.9323349972268442,
      "grad_norm": 0.04862720146775246,
      "learning_rate": 0.0001182716049382716,
      "loss": 0.5842,
      "step": 871
    },
    {
      "epoch": 1.9345535219079313,
      "grad_norm": 0.054187942296266556,
      "learning_rate": 0.00011802469135802468,
      "loss": 0.6265,
      "step": 872
    },
    {
      "epoch": 1.9367720465890184,
      "grad_norm": 0.04381798580288887,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.5187,
      "step": 873
    },
    {
      "epoch": 1.9389905712701054,
      "grad_norm": 0.04804033041000366,
      "learning_rate": 0.00011753086419753086,
      "loss": 0.621,
      "step": 874
    },
    {
      "epoch": 1.9412090959511925,
      "grad_norm": 0.047313276678323746,
      "learning_rate": 0.00011728395061728394,
      "loss": 0.6507,
      "step": 875
    },
    {
      "epoch": 1.9434276206322796,
      "grad_norm": 0.05348888784646988,
      "learning_rate": 0.00011703703703703702,
      "loss": 0.5533,
      "step": 876
    },
    {
      "epoch": 1.9456461453133667,
      "grad_norm": 0.05236828699707985,
      "learning_rate": 0.0001167901234567901,
      "loss": 0.4672,
      "step": 877
    },
    {
      "epoch": 1.9478646699944537,
      "grad_norm": 0.03475113585591316,
      "learning_rate": 0.00011654320987654319,
      "loss": 0.5282,
      "step": 878
    },
    {
      "epoch": 1.9500831946755408,
      "grad_norm": 0.044068362563848495,
      "learning_rate": 0.0001162962962962963,
      "loss": 0.6184,
      "step": 879
    },
    {
      "epoch": 1.9523017193566279,
      "grad_norm": 0.032331328839063644,
      "learning_rate": 0.00011604938271604938,
      "loss": 0.5171,
      "step": 880
    },
    {
      "epoch": 1.954520244037715,
      "grad_norm": 0.04091711714863777,
      "learning_rate": 0.00011580246913580246,
      "loss": 0.6441,
      "step": 881
    },
    {
      "epoch": 1.956738768718802,
      "grad_norm": 0.036669787019491196,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.5893,
      "step": 882
    },
    {
      "epoch": 1.958957293399889,
      "grad_norm": 0.041564103215932846,
      "learning_rate": 0.00011530864197530863,
      "loss": 0.6908,
      "step": 883
    },
    {
      "epoch": 1.9611758180809762,
      "grad_norm": 0.05795789510011673,
      "learning_rate": 0.00011506172839506172,
      "loss": 0.5791,
      "step": 884
    },
    {
      "epoch": 1.9633943427620633,
      "grad_norm": 0.04160236567258835,
      "learning_rate": 0.0001148148148148148,
      "loss": 0.6235,
      "step": 885
    },
    {
      "epoch": 1.9656128674431503,
      "grad_norm": 0.04947850480675697,
      "learning_rate": 0.00011456790123456789,
      "loss": 0.6072,
      "step": 886
    },
    {
      "epoch": 1.9678313921242374,
      "grad_norm": 0.05463534966111183,
      "learning_rate": 0.00011432098765432099,
      "loss": 0.5984,
      "step": 887
    },
    {
      "epoch": 1.9700499168053245,
      "grad_norm": 0.053235623985528946,
      "learning_rate": 0.00011407407407407407,
      "loss": 0.5788,
      "step": 888
    },
    {
      "epoch": 1.9722684414864116,
      "grad_norm": 0.03631250560283661,
      "learning_rate": 0.00011382716049382715,
      "loss": 0.5273,
      "step": 889
    },
    {
      "epoch": 1.9744869661674986,
      "grad_norm": 0.037177424877882004,
      "learning_rate": 0.00011358024691358023,
      "loss": 0.5107,
      "step": 890
    },
    {
      "epoch": 1.9767054908485857,
      "grad_norm": 0.05644045025110245,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.6884,
      "step": 891
    },
    {
      "epoch": 1.9789240155296728,
      "grad_norm": 0.036368608474731445,
      "learning_rate": 0.00011308641975308641,
      "loss": 0.4648,
      "step": 892
    },
    {
      "epoch": 1.9811425402107599,
      "grad_norm": 0.04657099395990372,
      "learning_rate": 0.00011283950617283949,
      "loss": 0.5458,
      "step": 893
    },
    {
      "epoch": 1.983361064891847,
      "grad_norm": 0.037454064935445786,
      "learning_rate": 0.00011259259259259259,
      "loss": 0.518,
      "step": 894
    },
    {
      "epoch": 1.985579589572934,
      "grad_norm": 0.03885742649435997,
      "learning_rate": 0.00011234567901234567,
      "loss": 0.6225,
      "step": 895
    },
    {
      "epoch": 1.987798114254021,
      "grad_norm": 0.04946479573845863,
      "learning_rate": 0.00011209876543209876,
      "loss": 0.6023,
      "step": 896
    },
    {
      "epoch": 1.9900166389351082,
      "grad_norm": 0.04252426326274872,
      "learning_rate": 0.00011185185185185184,
      "loss": 0.4941,
      "step": 897
    },
    {
      "epoch": 1.9922351636161952,
      "grad_norm": 0.03855001553893089,
      "learning_rate": 0.00011160493827160493,
      "loss": 0.4803,
      "step": 898
    },
    {
      "epoch": 1.9944536882972823,
      "grad_norm": 0.03618844971060753,
      "learning_rate": 0.00011135802469135801,
      "loss": 0.5881,
      "step": 899
    },
    {
      "epoch": 1.9966722129783694,
      "grad_norm": 0.0531810000538826,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.6604,
      "step": 900
    },
    {
      "epoch": 1.9988907376594565,
      "grad_norm": 0.046540580689907074,
      "learning_rate": 0.0001108641975308642,
      "loss": 0.5302,
      "step": 901
    },
    {
      "epoch": 2.0011092623405435,
      "grad_norm": 0.03842596337199211,
      "learning_rate": 0.00011061728395061728,
      "loss": 0.6209,
      "step": 902
    },
    {
      "epoch": 2.0033277870216306,
      "grad_norm": 0.0435018464922905,
      "learning_rate": 0.00011037037037037036,
      "loss": 0.4948,
      "step": 903
    },
    {
      "epoch": 2.0055463117027177,
      "grad_norm": 0.05771490931510925,
      "learning_rate": 0.00011012345679012344,
      "loss": 0.6608,
      "step": 904
    },
    {
      "epoch": 2.0077648363838048,
      "grad_norm": 0.06311209499835968,
      "learning_rate": 0.00010987654320987654,
      "loss": 0.6422,
      "step": 905
    },
    {
      "epoch": 2.009983361064892,
      "grad_norm": 0.06097199767827988,
      "learning_rate": 0.00010962962962962962,
      "loss": 0.5371,
      "step": 906
    },
    {
      "epoch": 2.012201885745979,
      "grad_norm": 0.04281490296125412,
      "learning_rate": 0.0001093827160493827,
      "loss": 0.5773,
      "step": 907
    },
    {
      "epoch": 2.014420410427066,
      "grad_norm": 0.04118552803993225,
      "learning_rate": 0.00010913580246913578,
      "loss": 0.611,
      "step": 908
    },
    {
      "epoch": 2.016638935108153,
      "grad_norm": 0.046561989933252335,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.6718,
      "step": 909
    },
    {
      "epoch": 2.01885745978924,
      "grad_norm": 0.057292282581329346,
      "learning_rate": 0.00010864197530864197,
      "loss": 0.6127,
      "step": 910
    },
    {
      "epoch": 2.021075984470327,
      "grad_norm": 0.04069270193576813,
      "learning_rate": 0.00010839506172839505,
      "loss": 0.6025,
      "step": 911
    },
    {
      "epoch": 2.0232945091514143,
      "grad_norm": 0.054237090051174164,
      "learning_rate": 0.00010814814814814814,
      "loss": 0.5396,
      "step": 912
    },
    {
      "epoch": 2.0255130338325014,
      "grad_norm": 0.052810315042734146,
      "learning_rate": 0.00010790123456790122,
      "loss": 0.4678,
      "step": 913
    },
    {
      "epoch": 2.0277315585135884,
      "grad_norm": 0.038033995777368546,
      "learning_rate": 0.00010765432098765431,
      "loss": 0.6086,
      "step": 914
    },
    {
      "epoch": 2.0299500831946755,
      "grad_norm": 0.07885255664587021,
      "learning_rate": 0.0001074074074074074,
      "loss": 0.5578,
      "step": 915
    },
    {
      "epoch": 2.0321686078757626,
      "grad_norm": 0.034748900681734085,
      "learning_rate": 0.00010716049382716048,
      "loss": 0.4578,
      "step": 916
    },
    {
      "epoch": 2.0343871325568497,
      "grad_norm": 0.10945303738117218,
      "learning_rate": 0.00010691358024691357,
      "loss": 0.5385,
      "step": 917
    },
    {
      "epoch": 2.0366056572379367,
      "grad_norm": 0.04476090893149376,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.5192,
      "step": 918
    },
    {
      "epoch": 2.038824181919024,
      "grad_norm": 0.04102242738008499,
      "learning_rate": 0.00010641975308641975,
      "loss": 0.7195,
      "step": 919
    },
    {
      "epoch": 2.041042706600111,
      "grad_norm": 0.03709735721349716,
      "learning_rate": 0.00010617283950617283,
      "loss": 0.4672,
      "step": 920
    },
    {
      "epoch": 2.043261231281198,
      "grad_norm": 0.0347595140337944,
      "learning_rate": 0.00010592592592592591,
      "loss": 0.5155,
      "step": 921
    },
    {
      "epoch": 2.045479755962285,
      "grad_norm": 0.03879978880286217,
      "learning_rate": 0.00010567901234567899,
      "loss": 0.5704,
      "step": 922
    },
    {
      "epoch": 2.047698280643372,
      "grad_norm": 0.03482053056359291,
      "learning_rate": 0.00010543209876543209,
      "loss": 0.4886,
      "step": 923
    },
    {
      "epoch": 2.049916805324459,
      "grad_norm": 0.04517371580004692,
      "learning_rate": 0.00010518518518518518,
      "loss": 0.4753,
      "step": 924
    },
    {
      "epoch": 2.0521353300055463,
      "grad_norm": 0.040182117372751236,
      "learning_rate": 0.00010493827160493826,
      "loss": 0.5035,
      "step": 925
    },
    {
      "epoch": 2.0543538546866333,
      "grad_norm": 0.03691491112112999,
      "learning_rate": 0.00010469135802469135,
      "loss": 0.6217,
      "step": 926
    },
    {
      "epoch": 2.0565723793677204,
      "grad_norm": 0.04582975059747696,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.6259,
      "step": 927
    },
    {
      "epoch": 2.0587909040488075,
      "grad_norm": 0.048866018652915955,
      "learning_rate": 0.00010419753086419752,
      "loss": 0.4704,
      "step": 928
    },
    {
      "epoch": 2.0610094287298946,
      "grad_norm": 0.04075740650296211,
      "learning_rate": 0.0001039506172839506,
      "loss": 0.6828,
      "step": 929
    },
    {
      "epoch": 2.0632279534109816,
      "grad_norm": 0.04279666021466255,
      "learning_rate": 0.00010370370370370369,
      "loss": 0.5505,
      "step": 930
    },
    {
      "epoch": 2.0654464780920687,
      "grad_norm": 0.03910258784890175,
      "learning_rate": 0.00010345679012345677,
      "loss": 0.7149,
      "step": 931
    },
    {
      "epoch": 2.067665002773156,
      "grad_norm": 0.0395696647465229,
      "learning_rate": 0.00010320987654320988,
      "loss": 0.4687,
      "step": 932
    },
    {
      "epoch": 2.069883527454243,
      "grad_norm": 0.04156024754047394,
      "learning_rate": 0.00010296296296296296,
      "loss": 0.6462,
      "step": 933
    },
    {
      "epoch": 2.07210205213533,
      "grad_norm": 0.04314883425831795,
      "learning_rate": 0.00010271604938271604,
      "loss": 0.6844,
      "step": 934
    },
    {
      "epoch": 2.074320576816417,
      "grad_norm": 0.041721161454916,
      "learning_rate": 0.00010246913580246912,
      "loss": 0.558,
      "step": 935
    },
    {
      "epoch": 2.076539101497504,
      "grad_norm": 0.04575420171022415,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.7513,
      "step": 936
    },
    {
      "epoch": 2.078757626178591,
      "grad_norm": 0.047548551112413406,
      "learning_rate": 0.0001019753086419753,
      "loss": 0.6336,
      "step": 937
    },
    {
      "epoch": 2.0809761508596782,
      "grad_norm": 0.04874306917190552,
      "learning_rate": 0.00010172839506172838,
      "loss": 0.4369,
      "step": 938
    },
    {
      "epoch": 2.0831946755407653,
      "grad_norm": 0.04517265036702156,
      "learning_rate": 0.00010148148148148147,
      "loss": 0.5553,
      "step": 939
    },
    {
      "epoch": 2.0854132002218524,
      "grad_norm": 0.03918233886361122,
      "learning_rate": 0.00010123456790123457,
      "loss": 0.5002,
      "step": 940
    },
    {
      "epoch": 2.0876317249029395,
      "grad_norm": 0.062212228775024414,
      "learning_rate": 0.00010098765432098765,
      "loss": 0.6709,
      "step": 941
    },
    {
      "epoch": 2.0898502495840265,
      "grad_norm": 0.047521647065877914,
      "learning_rate": 0.00010074074074074073,
      "loss": 0.6245,
      "step": 942
    },
    {
      "epoch": 2.0920687742651136,
      "grad_norm": 0.03871537744998932,
      "learning_rate": 0.00010049382716049381,
      "loss": 0.5872,
      "step": 943
    },
    {
      "epoch": 2.0942872989462007,
      "grad_norm": 0.035434190183877945,
      "learning_rate": 0.0001002469135802469,
      "loss": 0.525,
      "step": 944
    },
    {
      "epoch": 2.0965058236272878,
      "grad_norm": 0.03723178803920746,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.7193,
      "step": 945
    },
    {
      "epoch": 2.098724348308375,
      "grad_norm": 0.048287343233823776,
      "learning_rate": 9.975308641975307e-05,
      "loss": 0.6342,
      "step": 946
    },
    {
      "epoch": 2.100942872989462,
      "grad_norm": 0.043956004083156586,
      "learning_rate": 9.950617283950617e-05,
      "loss": 0.5647,
      "step": 947
    },
    {
      "epoch": 2.103161397670549,
      "grad_norm": 0.05077500268816948,
      "learning_rate": 9.925925925925925e-05,
      "loss": 0.5798,
      "step": 948
    },
    {
      "epoch": 2.105379922351636,
      "grad_norm": 0.035376980900764465,
      "learning_rate": 9.901234567901234e-05,
      "loss": 0.5213,
      "step": 949
    },
    {
      "epoch": 2.107598447032723,
      "grad_norm": 0.04889523983001709,
      "learning_rate": 9.876543209876543e-05,
      "loss": 0.5979,
      "step": 950
    },
    {
      "epoch": 2.10981697171381,
      "grad_norm": 0.046519458293914795,
      "learning_rate": 9.851851851851851e-05,
      "loss": 0.6326,
      "step": 951
    },
    {
      "epoch": 2.1120354963948973,
      "grad_norm": 0.03622965142130852,
      "learning_rate": 9.827160493827159e-05,
      "loss": 0.5006,
      "step": 952
    },
    {
      "epoch": 2.1142540210759844,
      "grad_norm": 0.06946947425603867,
      "learning_rate": 9.802469135802467e-05,
      "loss": 0.5952,
      "step": 953
    },
    {
      "epoch": 2.1164725457570714,
      "grad_norm": 0.038505807518959045,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.6425,
      "step": 954
    },
    {
      "epoch": 2.1186910704381585,
      "grad_norm": 0.048425644636154175,
      "learning_rate": 9.753086419753086e-05,
      "loss": 0.4956,
      "step": 955
    },
    {
      "epoch": 2.1209095951192456,
      "grad_norm": 0.05792173370718956,
      "learning_rate": 9.728395061728394e-05,
      "loss": 0.6387,
      "step": 956
    },
    {
      "epoch": 2.1231281198003327,
      "grad_norm": 0.04052269458770752,
      "learning_rate": 9.703703703703702e-05,
      "loss": 0.7024,
      "step": 957
    },
    {
      "epoch": 2.1253466444814197,
      "grad_norm": 0.040704745799303055,
      "learning_rate": 9.679012345679012e-05,
      "loss": 0.6304,
      "step": 958
    },
    {
      "epoch": 2.127565169162507,
      "grad_norm": 0.04877621680498123,
      "learning_rate": 9.65432098765432e-05,
      "loss": 0.5615,
      "step": 959
    },
    {
      "epoch": 2.129783693843594,
      "grad_norm": 0.06623969227075577,
      "learning_rate": 9.629629629629628e-05,
      "loss": 0.851,
      "step": 960
    },
    {
      "epoch": 2.132002218524681,
      "grad_norm": 0.047923821955919266,
      "learning_rate": 9.604938271604936e-05,
      "loss": 0.474,
      "step": 961
    },
    {
      "epoch": 2.134220743205768,
      "grad_norm": 0.06202734634280205,
      "learning_rate": 9.580246913580247e-05,
      "loss": 0.8462,
      "step": 962
    },
    {
      "epoch": 2.136439267886855,
      "grad_norm": 0.05249187722802162,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.6458,
      "step": 963
    },
    {
      "epoch": 2.138657792567942,
      "grad_norm": 0.04402559995651245,
      "learning_rate": 9.530864197530864e-05,
      "loss": 0.6105,
      "step": 964
    },
    {
      "epoch": 2.1408763172490293,
      "grad_norm": 0.05078354850411415,
      "learning_rate": 9.506172839506172e-05,
      "loss": 0.6583,
      "step": 965
    },
    {
      "epoch": 2.1430948419301163,
      "grad_norm": 0.04573201760649681,
      "learning_rate": 9.48148148148148e-05,
      "loss": 0.6684,
      "step": 966
    },
    {
      "epoch": 2.1453133666112034,
      "grad_norm": 0.03845048323273659,
      "learning_rate": 9.45679012345679e-05,
      "loss": 0.7102,
      "step": 967
    },
    {
      "epoch": 2.1475318912922905,
      "grad_norm": 0.034968942403793335,
      "learning_rate": 9.432098765432098e-05,
      "loss": 0.5046,
      "step": 968
    },
    {
      "epoch": 2.1497504159733776,
      "grad_norm": 0.03501512482762337,
      "learning_rate": 9.407407407407407e-05,
      "loss": 0.6369,
      "step": 969
    },
    {
      "epoch": 2.1519689406544646,
      "grad_norm": 0.04195564240217209,
      "learning_rate": 9.382716049382715e-05,
      "loss": 0.6742,
      "step": 970
    },
    {
      "epoch": 2.1541874653355517,
      "grad_norm": 0.07725167274475098,
      "learning_rate": 9.358024691358025e-05,
      "loss": 0.4538,
      "step": 971
    },
    {
      "epoch": 2.156405990016639,
      "grad_norm": 0.03606437146663666,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.4477,
      "step": 972
    },
    {
      "epoch": 2.158624514697726,
      "grad_norm": 0.037189386785030365,
      "learning_rate": 9.308641975308641e-05,
      "loss": 0.5911,
      "step": 973
    },
    {
      "epoch": 2.160843039378813,
      "grad_norm": 0.04275606945157051,
      "learning_rate": 9.283950617283949e-05,
      "loss": 0.6017,
      "step": 974
    },
    {
      "epoch": 2.1630615640599,
      "grad_norm": 0.035836249589920044,
      "learning_rate": 9.259259259259257e-05,
      "loss": 0.5703,
      "step": 975
    },
    {
      "epoch": 2.165280088740987,
      "grad_norm": 0.04088326171040535,
      "learning_rate": 9.234567901234567e-05,
      "loss": 0.5583,
      "step": 976
    },
    {
      "epoch": 2.167498613422074,
      "grad_norm": 0.030466226860880852,
      "learning_rate": 9.209876543209876e-05,
      "loss": 0.3768,
      "step": 977
    },
    {
      "epoch": 2.1697171381031612,
      "grad_norm": 0.04287552461028099,
      "learning_rate": 9.185185185185185e-05,
      "loss": 0.5495,
      "step": 978
    },
    {
      "epoch": 2.1719356627842483,
      "grad_norm": 0.0533541664481163,
      "learning_rate": 9.160493827160493e-05,
      "loss": 0.9231,
      "step": 979
    },
    {
      "epoch": 2.1741541874653354,
      "grad_norm": 0.051693834364414215,
      "learning_rate": 9.135802469135802e-05,
      "loss": 0.6359,
      "step": 980
    },
    {
      "epoch": 2.1763727121464225,
      "grad_norm": 0.04403116926550865,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.456,
      "step": 981
    },
    {
      "epoch": 2.1785912368275095,
      "grad_norm": 0.054048340767621994,
      "learning_rate": 9.086419753086419e-05,
      "loss": 0.5172,
      "step": 982
    },
    {
      "epoch": 2.1808097615085966,
      "grad_norm": 0.07537399232387543,
      "learning_rate": 9.061728395061727e-05,
      "loss": 0.7758,
      "step": 983
    },
    {
      "epoch": 2.1830282861896837,
      "grad_norm": 0.03217058256268501,
      "learning_rate": 9.037037037037038e-05,
      "loss": 0.5199,
      "step": 984
    },
    {
      "epoch": 2.1852468108707708,
      "grad_norm": 0.04562436789274216,
      "learning_rate": 9.012345679012346e-05,
      "loss": 0.563,
      "step": 985
    },
    {
      "epoch": 2.187465335551858,
      "grad_norm": 0.03595530241727829,
      "learning_rate": 8.987654320987654e-05,
      "loss": 0.5712,
      "step": 986
    },
    {
      "epoch": 2.189683860232945,
      "grad_norm": 0.06466427445411682,
      "learning_rate": 8.962962962962962e-05,
      "loss": 0.3933,
      "step": 987
    },
    {
      "epoch": 2.191902384914032,
      "grad_norm": 0.05376742407679558,
      "learning_rate": 8.93827160493827e-05,
      "loss": 0.5676,
      "step": 988
    },
    {
      "epoch": 2.194120909595119,
      "grad_norm": 0.0453377328813076,
      "learning_rate": 8.91358024691358e-05,
      "loss": 0.4984,
      "step": 989
    },
    {
      "epoch": 2.196339434276206,
      "grad_norm": 0.04033369943499565,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.5528,
      "step": 990
    },
    {
      "epoch": 2.1985579589572932,
      "grad_norm": 0.03792077675461769,
      "learning_rate": 8.864197530864196e-05,
      "loss": 0.5399,
      "step": 991
    },
    {
      "epoch": 2.2007764836383803,
      "grad_norm": 0.040184441953897476,
      "learning_rate": 8.839506172839506e-05,
      "loss": 0.4738,
      "step": 992
    },
    {
      "epoch": 2.2029950083194674,
      "grad_norm": 0.04826079681515694,
      "learning_rate": 8.814814814814815e-05,
      "loss": 0.9864,
      "step": 993
    },
    {
      "epoch": 2.2052135330005544,
      "grad_norm": 0.04012729600071907,
      "learning_rate": 8.790123456790123e-05,
      "loss": 0.4425,
      "step": 994
    },
    {
      "epoch": 2.2074320576816415,
      "grad_norm": 0.042340125888586044,
      "learning_rate": 8.765432098765431e-05,
      "loss": 0.4283,
      "step": 995
    },
    {
      "epoch": 2.2096505823627286,
      "grad_norm": 0.05345200374722481,
      "learning_rate": 8.74074074074074e-05,
      "loss": 0.6981,
      "step": 996
    },
    {
      "epoch": 2.2118691070438157,
      "grad_norm": 0.04511335864663124,
      "learning_rate": 8.716049382716048e-05,
      "loss": 0.5142,
      "step": 997
    },
    {
      "epoch": 2.2140876317249027,
      "grad_norm": 0.041533827781677246,
      "learning_rate": 8.691358024691357e-05,
      "loss": 0.6515,
      "step": 998
    },
    {
      "epoch": 2.21630615640599,
      "grad_norm": 0.04999273270368576,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.4537,
      "step": 999
    },
    {
      "epoch": 2.218524681087077,
      "grad_norm": 0.06083151698112488,
      "learning_rate": 8.641975308641975e-05,
      "loss": 0.589,
      "step": 1000
    },
    {
      "epoch": 2.220743205768164,
      "grad_norm": 0.05858886241912842,
      "learning_rate": 8.617283950617283e-05,
      "loss": 0.6339,
      "step": 1001
    },
    {
      "epoch": 2.222961730449251,
      "grad_norm": 0.03843481093645096,
      "learning_rate": 8.592592592592593e-05,
      "loss": 0.5983,
      "step": 1002
    },
    {
      "epoch": 2.225180255130338,
      "grad_norm": 0.035686057060956955,
      "learning_rate": 8.567901234567901e-05,
      "loss": 0.5206,
      "step": 1003
    },
    {
      "epoch": 2.227398779811425,
      "grad_norm": 0.06216483935713768,
      "learning_rate": 8.543209876543209e-05,
      "loss": 0.5241,
      "step": 1004
    },
    {
      "epoch": 2.2296173044925123,
      "grad_norm": 0.06830698996782303,
      "learning_rate": 8.518518518518517e-05,
      "loss": 0.5416,
      "step": 1005
    },
    {
      "epoch": 2.2318358291735994,
      "grad_norm": 0.045379605144262314,
      "learning_rate": 8.493827160493825e-05,
      "loss": 0.6191,
      "step": 1006
    },
    {
      "epoch": 2.2340543538546864,
      "grad_norm": 0.046395085752010345,
      "learning_rate": 8.469135802469136e-05,
      "loss": 0.4947,
      "step": 1007
    },
    {
      "epoch": 2.2362728785357735,
      "grad_norm": 0.05502278357744217,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.6531,
      "step": 1008
    },
    {
      "epoch": 2.2384914032168606,
      "grad_norm": 0.03305497393012047,
      "learning_rate": 8.419753086419752e-05,
      "loss": 0.597,
      "step": 1009
    },
    {
      "epoch": 2.2407099278979477,
      "grad_norm": 0.0502849742770195,
      "learning_rate": 8.39506172839506e-05,
      "loss": 0.453,
      "step": 1010
    },
    {
      "epoch": 2.2429284525790347,
      "grad_norm": 0.04740210622549057,
      "learning_rate": 8.37037037037037e-05,
      "loss": 0.5545,
      "step": 1011
    },
    {
      "epoch": 2.245146977260122,
      "grad_norm": 0.03833049535751343,
      "learning_rate": 8.345679012345678e-05,
      "loss": 0.4932,
      "step": 1012
    },
    {
      "epoch": 2.247365501941209,
      "grad_norm": 0.06829188764095306,
      "learning_rate": 8.320987654320986e-05,
      "loss": 0.5887,
      "step": 1013
    },
    {
      "epoch": 2.249584026622296,
      "grad_norm": 0.05020642653107643,
      "learning_rate": 8.296296296296295e-05,
      "loss": 0.6925,
      "step": 1014
    },
    {
      "epoch": 2.251802551303383,
      "grad_norm": 0.035927675664424896,
      "learning_rate": 8.271604938271605e-05,
      "loss": 0.674,
      "step": 1015
    },
    {
      "epoch": 2.25402107598447,
      "grad_norm": 0.04108450189232826,
      "learning_rate": 8.246913580246914e-05,
      "loss": 0.6315,
      "step": 1016
    },
    {
      "epoch": 2.256239600665557,
      "grad_norm": 0.0487007237970829,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.7524,
      "step": 1017
    },
    {
      "epoch": 2.2584581253466443,
      "grad_norm": 0.05985472723841667,
      "learning_rate": 8.19753086419753e-05,
      "loss": 0.7626,
      "step": 1018
    },
    {
      "epoch": 2.2606766500277313,
      "grad_norm": 0.037847504019737244,
      "learning_rate": 8.172839506172838e-05,
      "loss": 0.6937,
      "step": 1019
    },
    {
      "epoch": 2.2628951747088184,
      "grad_norm": 0.04677148535847664,
      "learning_rate": 8.148148148148148e-05,
      "loss": 0.6935,
      "step": 1020
    },
    {
      "epoch": 2.2651136993899055,
      "grad_norm": 0.07024740427732468,
      "learning_rate": 8.123456790123456e-05,
      "loss": 0.6643,
      "step": 1021
    },
    {
      "epoch": 2.2673322240709926,
      "grad_norm": 0.06649952381849289,
      "learning_rate": 8.098765432098765e-05,
      "loss": 0.5444,
      "step": 1022
    },
    {
      "epoch": 2.2695507487520796,
      "grad_norm": 0.049044687300920486,
      "learning_rate": 8.074074074074073e-05,
      "loss": 0.6385,
      "step": 1023
    },
    {
      "epoch": 2.2717692734331667,
      "grad_norm": 0.05241164565086365,
      "learning_rate": 8.049382716049383e-05,
      "loss": 0.6628,
      "step": 1024
    },
    {
      "epoch": 2.2739877981142538,
      "grad_norm": 0.03445402905344963,
      "learning_rate": 8.024691358024691e-05,
      "loss": 0.604,
      "step": 1025
    },
    {
      "epoch": 2.276206322795341,
      "grad_norm": 0.06771976500749588,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.614,
      "step": 1026
    },
    {
      "epoch": 2.278424847476428,
      "grad_norm": 0.10511960834264755,
      "learning_rate": 7.975308641975307e-05,
      "loss": 0.5576,
      "step": 1027
    },
    {
      "epoch": 2.280643372157515,
      "grad_norm": 0.05881739780306816,
      "learning_rate": 7.950617283950616e-05,
      "loss": 0.4848,
      "step": 1028
    },
    {
      "epoch": 2.2828618968386025,
      "grad_norm": 0.037616219371557236,
      "learning_rate": 7.925925925925925e-05,
      "loss": 0.5778,
      "step": 1029
    },
    {
      "epoch": 2.2850804215196896,
      "grad_norm": 0.07373534142971039,
      "learning_rate": 7.901234567901235e-05,
      "loss": 0.6883,
      "step": 1030
    },
    {
      "epoch": 2.2872989462007767,
      "grad_norm": 0.04770660027861595,
      "learning_rate": 7.876543209876543e-05,
      "loss": 0.6981,
      "step": 1031
    },
    {
      "epoch": 2.2895174708818637,
      "grad_norm": 0.037155069410800934,
      "learning_rate": 7.851851851851851e-05,
      "loss": 0.615,
      "step": 1032
    },
    {
      "epoch": 2.291735995562951,
      "grad_norm": 0.0474337600171566,
      "learning_rate": 7.82716049382716e-05,
      "loss": 0.4782,
      "step": 1033
    },
    {
      "epoch": 2.293954520244038,
      "grad_norm": 0.05592214688658714,
      "learning_rate": 7.802469135802469e-05,
      "loss": 0.6584,
      "step": 1034
    },
    {
      "epoch": 2.296173044925125,
      "grad_norm": 0.03420605883002281,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.4572,
      "step": 1035
    },
    {
      "epoch": 2.298391569606212,
      "grad_norm": 0.0462031289935112,
      "learning_rate": 7.753086419753085e-05,
      "loss": 0.5716,
      "step": 1036
    },
    {
      "epoch": 2.300610094287299,
      "grad_norm": 0.038128022104501724,
      "learning_rate": 7.728395061728396e-05,
      "loss": 0.48,
      "step": 1037
    },
    {
      "epoch": 2.302828618968386,
      "grad_norm": 0.03226501867175102,
      "learning_rate": 7.703703703703704e-05,
      "loss": 0.6333,
      "step": 1038
    },
    {
      "epoch": 2.3050471436494733,
      "grad_norm": 0.04452360421419144,
      "learning_rate": 7.679012345679012e-05,
      "loss": 0.591,
      "step": 1039
    },
    {
      "epoch": 2.3072656683305603,
      "grad_norm": 0.04380777105689049,
      "learning_rate": 7.65432098765432e-05,
      "loss": 0.5631,
      "step": 1040
    },
    {
      "epoch": 2.3094841930116474,
      "grad_norm": 0.05589490383863449,
      "learning_rate": 7.629629629629628e-05,
      "loss": 0.6814,
      "step": 1041
    },
    {
      "epoch": 2.3117027176927345,
      "grad_norm": 0.03977074474096298,
      "learning_rate": 7.604938271604937e-05,
      "loss": 0.5611,
      "step": 1042
    },
    {
      "epoch": 2.3139212423738216,
      "grad_norm": 0.056011974811553955,
      "learning_rate": 7.580246913580246e-05,
      "loss": 0.6049,
      "step": 1043
    },
    {
      "epoch": 2.3161397670549086,
      "grad_norm": 0.03692979738116264,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.5612,
      "step": 1044
    },
    {
      "epoch": 2.3183582917359957,
      "grad_norm": 0.05938183516263962,
      "learning_rate": 7.530864197530864e-05,
      "loss": 0.6598,
      "step": 1045
    },
    {
      "epoch": 2.320576816417083,
      "grad_norm": 0.04410908371210098,
      "learning_rate": 7.506172839506173e-05,
      "loss": 0.6726,
      "step": 1046
    },
    {
      "epoch": 2.32279534109817,
      "grad_norm": 0.047143902629613876,
      "learning_rate": 7.481481481481481e-05,
      "loss": 0.5246,
      "step": 1047
    },
    {
      "epoch": 2.325013865779257,
      "grad_norm": 0.03511461615562439,
      "learning_rate": 7.45679012345679e-05,
      "loss": 0.5641,
      "step": 1048
    },
    {
      "epoch": 2.327232390460344,
      "grad_norm": 0.10739941149950027,
      "learning_rate": 7.432098765432098e-05,
      "loss": 0.6458,
      "step": 1049
    },
    {
      "epoch": 2.329450915141431,
      "grad_norm": 0.04435363784432411,
      "learning_rate": 7.407407407407406e-05,
      "loss": 0.6077,
      "step": 1050
    },
    {
      "epoch": 2.331669439822518,
      "grad_norm": 0.07181151211261749,
      "learning_rate": 7.382716049382715e-05,
      "loss": 0.6099,
      "step": 1051
    },
    {
      "epoch": 2.3338879645036053,
      "grad_norm": 0.03596745431423187,
      "learning_rate": 7.358024691358024e-05,
      "loss": 0.4921,
      "step": 1052
    },
    {
      "epoch": 2.3361064891846923,
      "grad_norm": 0.07310967147350311,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.7241,
      "step": 1053
    },
    {
      "epoch": 2.3383250138657794,
      "grad_norm": 0.06407670676708221,
      "learning_rate": 7.308641975308641e-05,
      "loss": 0.6485,
      "step": 1054
    },
    {
      "epoch": 2.3405435385468665,
      "grad_norm": 0.048728350549936295,
      "learning_rate": 7.28395061728395e-05,
      "loss": 0.6286,
      "step": 1055
    },
    {
      "epoch": 2.3427620632279536,
      "grad_norm": 0.03892116621136665,
      "learning_rate": 7.259259259259259e-05,
      "loss": 0.6301,
      "step": 1056
    },
    {
      "epoch": 2.3449805879090406,
      "grad_norm": 0.050711289048194885,
      "learning_rate": 7.234567901234567e-05,
      "loss": 0.5772,
      "step": 1057
    },
    {
      "epoch": 2.3471991125901277,
      "grad_norm": 0.07137377560138702,
      "learning_rate": 7.209876543209877e-05,
      "loss": 0.5935,
      "step": 1058
    },
    {
      "epoch": 2.3494176372712148,
      "grad_norm": 0.035764068365097046,
      "learning_rate": 7.185185185185185e-05,
      "loss": 0.5725,
      "step": 1059
    },
    {
      "epoch": 2.351636161952302,
      "grad_norm": 0.03853100910782814,
      "learning_rate": 7.160493827160493e-05,
      "loss": 0.5075,
      "step": 1060
    },
    {
      "epoch": 2.353854686633389,
      "grad_norm": 0.03932284191250801,
      "learning_rate": 7.135802469135801e-05,
      "loss": 0.5324,
      "step": 1061
    },
    {
      "epoch": 2.356073211314476,
      "grad_norm": 0.03677055239677429,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.5779,
      "step": 1062
    },
    {
      "epoch": 2.358291735995563,
      "grad_norm": 0.03266173601150513,
      "learning_rate": 7.086419753086419e-05,
      "loss": 0.4891,
      "step": 1063
    },
    {
      "epoch": 2.36051026067665,
      "grad_norm": 0.04765373468399048,
      "learning_rate": 7.061728395061727e-05,
      "loss": 0.6248,
      "step": 1064
    },
    {
      "epoch": 2.3627287853577372,
      "grad_norm": 0.0494760200381279,
      "learning_rate": 7.037037037037036e-05,
      "loss": 0.5336,
      "step": 1065
    },
    {
      "epoch": 2.3649473100388243,
      "grad_norm": 0.02984657697379589,
      "learning_rate": 7.012345679012345e-05,
      "loss": 0.5167,
      "step": 1066
    },
    {
      "epoch": 2.3671658347199114,
      "grad_norm": 0.05170505866408348,
      "learning_rate": 6.987654320987654e-05,
      "loss": 0.6803,
      "step": 1067
    },
    {
      "epoch": 2.3693843594009985,
      "grad_norm": 0.05598672479391098,
      "learning_rate": 6.962962962962962e-05,
      "loss": 0.7547,
      "step": 1068
    },
    {
      "epoch": 2.3716028840820855,
      "grad_norm": 0.030783263966441154,
      "learning_rate": 6.938271604938272e-05,
      "loss": 0.4325,
      "step": 1069
    },
    {
      "epoch": 2.3738214087631726,
      "grad_norm": 0.054061245173215866,
      "learning_rate": 6.91358024691358e-05,
      "loss": 0.5373,
      "step": 1070
    },
    {
      "epoch": 2.3760399334442597,
      "grad_norm": 0.05096113309264183,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.7236,
      "step": 1071
    },
    {
      "epoch": 2.3782584581253468,
      "grad_norm": 0.03470573201775551,
      "learning_rate": 6.864197530864196e-05,
      "loss": 0.6409,
      "step": 1072
    },
    {
      "epoch": 2.380476982806434,
      "grad_norm": 0.04524906352162361,
      "learning_rate": 6.839506172839506e-05,
      "loss": 0.6695,
      "step": 1073
    },
    {
      "epoch": 2.382695507487521,
      "grad_norm": 0.042583104223012924,
      "learning_rate": 6.814814814814814e-05,
      "loss": 0.6238,
      "step": 1074
    },
    {
      "epoch": 2.384914032168608,
      "grad_norm": 0.08511609584093094,
      "learning_rate": 6.790123456790122e-05,
      "loss": 0.4686,
      "step": 1075
    },
    {
      "epoch": 2.387132556849695,
      "grad_norm": 0.05178077146410942,
      "learning_rate": 6.765432098765432e-05,
      "loss": 0.5263,
      "step": 1076
    },
    {
      "epoch": 2.389351081530782,
      "grad_norm": 0.04367220401763916,
      "learning_rate": 6.74074074074074e-05,
      "loss": 0.7811,
      "step": 1077
    },
    {
      "epoch": 2.391569606211869,
      "grad_norm": 0.07618217170238495,
      "learning_rate": 6.716049382716049e-05,
      "loss": 0.8423,
      "step": 1078
    },
    {
      "epoch": 2.3937881308929563,
      "grad_norm": 0.06087173894047737,
      "learning_rate": 6.691358024691357e-05,
      "loss": 0.6306,
      "step": 1079
    },
    {
      "epoch": 2.3960066555740434,
      "grad_norm": 0.03554554656147957,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.5626,
      "step": 1080
    },
    {
      "epoch": 2.3982251802551304,
      "grad_norm": 0.045016031712293625,
      "learning_rate": 6.641975308641975e-05,
      "loss": 0.5935,
      "step": 1081
    },
    {
      "epoch": 2.4004437049362175,
      "grad_norm": 0.03879620134830475,
      "learning_rate": 6.617283950617283e-05,
      "loss": 0.542,
      "step": 1082
    },
    {
      "epoch": 2.4026622296173046,
      "grad_norm": 0.05213743820786476,
      "learning_rate": 6.592592592592591e-05,
      "loss": 0.5068,
      "step": 1083
    },
    {
      "epoch": 2.4048807542983917,
      "grad_norm": 0.044363513588905334,
      "learning_rate": 6.567901234567901e-05,
      "loss": 0.5524,
      "step": 1084
    },
    {
      "epoch": 2.4070992789794787,
      "grad_norm": 0.03913717344403267,
      "learning_rate": 6.543209876543209e-05,
      "loss": 0.4935,
      "step": 1085
    },
    {
      "epoch": 2.409317803660566,
      "grad_norm": 0.036936916410923004,
      "learning_rate": 6.518518518518517e-05,
      "loss": 0.5136,
      "step": 1086
    },
    {
      "epoch": 2.411536328341653,
      "grad_norm": 0.03922627493739128,
      "learning_rate": 6.493827160493827e-05,
      "loss": 0.4974,
      "step": 1087
    },
    {
      "epoch": 2.41375485302274,
      "grad_norm": 0.04165932908654213,
      "learning_rate": 6.469135802469135e-05,
      "loss": 0.4581,
      "step": 1088
    },
    {
      "epoch": 2.415973377703827,
      "grad_norm": 0.048920221626758575,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.4764,
      "step": 1089
    },
    {
      "epoch": 2.418191902384914,
      "grad_norm": 0.04519141465425491,
      "learning_rate": 6.419753086419753e-05,
      "loss": 0.5897,
      "step": 1090
    },
    {
      "epoch": 2.420410427066001,
      "grad_norm": 0.033349402248859406,
      "learning_rate": 6.395061728395061e-05,
      "loss": 0.5809,
      "step": 1091
    },
    {
      "epoch": 2.4226289517470883,
      "grad_norm": 0.047301311045885086,
      "learning_rate": 6.37037037037037e-05,
      "loss": 0.4635,
      "step": 1092
    },
    {
      "epoch": 2.4248474764281753,
      "grad_norm": 0.043773092329502106,
      "learning_rate": 6.345679012345678e-05,
      "loss": 0.5804,
      "step": 1093
    },
    {
      "epoch": 2.4270660011092624,
      "grad_norm": 0.05782271921634674,
      "learning_rate": 6.320987654320987e-05,
      "loss": 0.7331,
      "step": 1094
    },
    {
      "epoch": 2.4292845257903495,
      "grad_norm": 0.033670298755168915,
      "learning_rate": 6.296296296296295e-05,
      "loss": 0.4515,
      "step": 1095
    },
    {
      "epoch": 2.4315030504714366,
      "grad_norm": 0.03698200732469559,
      "learning_rate": 6.271604938271604e-05,
      "loss": 0.6,
      "step": 1096
    },
    {
      "epoch": 2.4337215751525236,
      "grad_norm": 0.04562830179929733,
      "learning_rate": 6.246913580246912e-05,
      "loss": 0.5608,
      "step": 1097
    },
    {
      "epoch": 2.4359400998336107,
      "grad_norm": 0.06514487415552139,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.6041,
      "step": 1098
    },
    {
      "epoch": 2.438158624514698,
      "grad_norm": 0.034314680844545364,
      "learning_rate": 6.19753086419753e-05,
      "loss": 0.5304,
      "step": 1099
    },
    {
      "epoch": 2.440377149195785,
      "grad_norm": 0.059299129992723465,
      "learning_rate": 6.17283950617284e-05,
      "loss": 0.5249,
      "step": 1100
    },
    {
      "epoch": 2.442595673876872,
      "grad_norm": 0.04950685054063797,
      "learning_rate": 6.148148148148148e-05,
      "loss": 0.5737,
      "step": 1101
    },
    {
      "epoch": 2.444814198557959,
      "grad_norm": 0.03410561755299568,
      "learning_rate": 6.123456790123456e-05,
      "loss": 0.5441,
      "step": 1102
    },
    {
      "epoch": 2.447032723239046,
      "grad_norm": 0.040159136056900024,
      "learning_rate": 6.0987654320987655e-05,
      "loss": 0.5859,
      "step": 1103
    },
    {
      "epoch": 2.449251247920133,
      "grad_norm": 0.04181082546710968,
      "learning_rate": 6.0740740740740736e-05,
      "loss": 0.4503,
      "step": 1104
    },
    {
      "epoch": 2.4514697726012202,
      "grad_norm": 0.03692084178328514,
      "learning_rate": 6.049382716049382e-05,
      "loss": 0.49,
      "step": 1105
    },
    {
      "epoch": 2.4536882972823073,
      "grad_norm": 0.03933589905500412,
      "learning_rate": 6.0246913580246906e-05,
      "loss": 0.5456,
      "step": 1106
    },
    {
      "epoch": 2.4559068219633944,
      "grad_norm": 0.053087152540683746,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.541,
      "step": 1107
    },
    {
      "epoch": 2.4581253466444815,
      "grad_norm": 0.03665045276284218,
      "learning_rate": 5.975308641975308e-05,
      "loss": 0.4868,
      "step": 1108
    },
    {
      "epoch": 2.4603438713255685,
      "grad_norm": 0.045733705163002014,
      "learning_rate": 5.9506172839506165e-05,
      "loss": 0.4333,
      "step": 1109
    },
    {
      "epoch": 2.4625623960066556,
      "grad_norm": 0.041406091302633286,
      "learning_rate": 5.925925925925925e-05,
      "loss": 0.5384,
      "step": 1110
    },
    {
      "epoch": 2.4647809206877427,
      "grad_norm": 0.04350060969591141,
      "learning_rate": 5.901234567901234e-05,
      "loss": 0.5348,
      "step": 1111
    },
    {
      "epoch": 2.4669994453688298,
      "grad_norm": 0.03200163319706917,
      "learning_rate": 5.876543209876543e-05,
      "loss": 0.4658,
      "step": 1112
    },
    {
      "epoch": 2.469217970049917,
      "grad_norm": 0.05263541638851166,
      "learning_rate": 5.851851851851851e-05,
      "loss": 0.5586,
      "step": 1113
    },
    {
      "epoch": 2.471436494731004,
      "grad_norm": 0.0404154509305954,
      "learning_rate": 5.827160493827159e-05,
      "loss": 0.5694,
      "step": 1114
    },
    {
      "epoch": 2.473655019412091,
      "grad_norm": 0.038837023079395294,
      "learning_rate": 5.802469135802469e-05,
      "loss": 0.51,
      "step": 1115
    },
    {
      "epoch": 2.475873544093178,
      "grad_norm": 0.06873560696840286,
      "learning_rate": 5.777777777777777e-05,
      "loss": 0.7588,
      "step": 1116
    },
    {
      "epoch": 2.478092068774265,
      "grad_norm": 0.042499661445617676,
      "learning_rate": 5.753086419753086e-05,
      "loss": 0.5495,
      "step": 1117
    },
    {
      "epoch": 2.480310593455352,
      "grad_norm": 0.04088939353823662,
      "learning_rate": 5.7283950617283946e-05,
      "loss": 0.5235,
      "step": 1118
    },
    {
      "epoch": 2.4825291181364393,
      "grad_norm": 0.03828609734773636,
      "learning_rate": 5.7037037037037035e-05,
      "loss": 0.5798,
      "step": 1119
    },
    {
      "epoch": 2.4847476428175264,
      "grad_norm": 0.05818233639001846,
      "learning_rate": 5.6790123456790116e-05,
      "loss": 0.6146,
      "step": 1120
    },
    {
      "epoch": 2.4869661674986134,
      "grad_norm": 0.04786087945103645,
      "learning_rate": 5.6543209876543205e-05,
      "loss": 0.5424,
      "step": 1121
    },
    {
      "epoch": 2.4891846921797005,
      "grad_norm": 0.04645780473947525,
      "learning_rate": 5.629629629629629e-05,
      "loss": 0.6642,
      "step": 1122
    },
    {
      "epoch": 2.4914032168607876,
      "grad_norm": 0.044215165078639984,
      "learning_rate": 5.604938271604938e-05,
      "loss": 0.595,
      "step": 1123
    },
    {
      "epoch": 2.4936217415418747,
      "grad_norm": 0.046210601925849915,
      "learning_rate": 5.580246913580246e-05,
      "loss": 0.6446,
      "step": 1124
    },
    {
      "epoch": 2.4958402662229617,
      "grad_norm": 0.033865753561258316,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.6187,
      "step": 1125
    },
    {
      "epoch": 2.498058790904049,
      "grad_norm": 0.04445701837539673,
      "learning_rate": 5.530864197530864e-05,
      "loss": 0.5842,
      "step": 1126
    },
    {
      "epoch": 2.500277315585136,
      "grad_norm": 0.041012953966856,
      "learning_rate": 5.506172839506172e-05,
      "loss": 0.4846,
      "step": 1127
    },
    {
      "epoch": 2.502495840266223,
      "grad_norm": 0.04022415727376938,
      "learning_rate": 5.481481481481481e-05,
      "loss": 0.5478,
      "step": 1128
    },
    {
      "epoch": 2.50471436494731,
      "grad_norm": 0.0493166409432888,
      "learning_rate": 5.456790123456789e-05,
      "loss": 0.6112,
      "step": 1129
    },
    {
      "epoch": 2.506932889628397,
      "grad_norm": 0.046915240585803986,
      "learning_rate": 5.4320987654320986e-05,
      "loss": 0.5151,
      "step": 1130
    },
    {
      "epoch": 2.509151414309484,
      "grad_norm": 0.052590250968933105,
      "learning_rate": 5.407407407407407e-05,
      "loss": 0.5514,
      "step": 1131
    },
    {
      "epoch": 2.5113699389905713,
      "grad_norm": 0.04393552988767624,
      "learning_rate": 5.3827160493827156e-05,
      "loss": 0.6526,
      "step": 1132
    },
    {
      "epoch": 2.5135884636716583,
      "grad_norm": 0.03863636776804924,
      "learning_rate": 5.358024691358024e-05,
      "loss": 0.4916,
      "step": 1133
    },
    {
      "epoch": 2.5158069883527454,
      "grad_norm": 0.0461195632815361,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.6713,
      "step": 1134
    },
    {
      "epoch": 2.5180255130338325,
      "grad_norm": 0.07105227559804916,
      "learning_rate": 5.3086419753086415e-05,
      "loss": 0.4187,
      "step": 1135
    },
    {
      "epoch": 2.5202440377149196,
      "grad_norm": 0.06375302374362946,
      "learning_rate": 5.2839506172839496e-05,
      "loss": 0.6636,
      "step": 1136
    },
    {
      "epoch": 2.5224625623960066,
      "grad_norm": 0.03883707895874977,
      "learning_rate": 5.259259259259259e-05,
      "loss": 0.5808,
      "step": 1137
    },
    {
      "epoch": 2.5246810870770937,
      "grad_norm": 0.038865018635988235,
      "learning_rate": 5.234567901234567e-05,
      "loss": 0.5451,
      "step": 1138
    },
    {
      "epoch": 2.526899611758181,
      "grad_norm": 0.05583466216921806,
      "learning_rate": 5.209876543209876e-05,
      "loss": 0.8092,
      "step": 1139
    },
    {
      "epoch": 2.529118136439268,
      "grad_norm": 0.040964554995298386,
      "learning_rate": 5.185185185185184e-05,
      "loss": 0.5344,
      "step": 1140
    },
    {
      "epoch": 2.531336661120355,
      "grad_norm": 0.05414204299449921,
      "learning_rate": 5.160493827160494e-05,
      "loss": 0.5067,
      "step": 1141
    },
    {
      "epoch": 2.533555185801442,
      "grad_norm": 0.04235865920782089,
      "learning_rate": 5.135802469135802e-05,
      "loss": 0.6309,
      "step": 1142
    },
    {
      "epoch": 2.535773710482529,
      "grad_norm": 0.048164498060941696,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.6318,
      "step": 1143
    },
    {
      "epoch": 2.537992235163616,
      "grad_norm": 0.06235666200518608,
      "learning_rate": 5.086419753086419e-05,
      "loss": 0.9454,
      "step": 1144
    },
    {
      "epoch": 2.5402107598447032,
      "grad_norm": 0.04797593504190445,
      "learning_rate": 5.0617283950617285e-05,
      "loss": 0.4991,
      "step": 1145
    },
    {
      "epoch": 2.5424292845257903,
      "grad_norm": 0.04417087510228157,
      "learning_rate": 5.0370370370370366e-05,
      "loss": 0.5738,
      "step": 1146
    },
    {
      "epoch": 2.5446478092068774,
      "grad_norm": 0.044092241674661636,
      "learning_rate": 5.012345679012345e-05,
      "loss": 0.8354,
      "step": 1147
    },
    {
      "epoch": 2.5468663338879645,
      "grad_norm": 0.06271953880786896,
      "learning_rate": 4.9876543209876536e-05,
      "loss": 0.6506,
      "step": 1148
    },
    {
      "epoch": 2.5490848585690515,
      "grad_norm": 0.048784974962472916,
      "learning_rate": 4.9629629629629625e-05,
      "loss": 0.6083,
      "step": 1149
    },
    {
      "epoch": 2.5513033832501386,
      "grad_norm": 0.06643613427877426,
      "learning_rate": 4.938271604938271e-05,
      "loss": 0.6037,
      "step": 1150
    },
    {
      "epoch": 2.5535219079312257,
      "grad_norm": 0.05064278841018677,
      "learning_rate": 4.9135802469135795e-05,
      "loss": 0.5944,
      "step": 1151
    },
    {
      "epoch": 2.5557404326123128,
      "grad_norm": 0.0397254079580307,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.5927,
      "step": 1152
    },
    {
      "epoch": 2.5579589572934,
      "grad_norm": 0.04077325388789177,
      "learning_rate": 4.864197530864197e-05,
      "loss": 0.4854,
      "step": 1153
    },
    {
      "epoch": 2.560177481974487,
      "grad_norm": 0.05955452844500542,
      "learning_rate": 4.839506172839506e-05,
      "loss": 0.5992,
      "step": 1154
    },
    {
      "epoch": 2.562396006655574,
      "grad_norm": 0.05283373221755028,
      "learning_rate": 4.814814814814814e-05,
      "loss": 0.6556,
      "step": 1155
    },
    {
      "epoch": 2.564614531336661,
      "grad_norm": 0.04814062640070915,
      "learning_rate": 4.7901234567901237e-05,
      "loss": 0.5941,
      "step": 1156
    },
    {
      "epoch": 2.566833056017748,
      "grad_norm": 0.042466599494218826,
      "learning_rate": 4.765432098765432e-05,
      "loss": 0.4389,
      "step": 1157
    },
    {
      "epoch": 2.569051580698835,
      "grad_norm": 0.037445127964019775,
      "learning_rate": 4.74074074074074e-05,
      "loss": 0.6236,
      "step": 1158
    },
    {
      "epoch": 2.5712701053799223,
      "grad_norm": 0.040361132472753525,
      "learning_rate": 4.716049382716049e-05,
      "loss": 0.5983,
      "step": 1159
    },
    {
      "epoch": 2.5734886300610094,
      "grad_norm": 0.041336093097925186,
      "learning_rate": 4.6913580246913576e-05,
      "loss": 0.589,
      "step": 1160
    },
    {
      "epoch": 2.5757071547420964,
      "grad_norm": 0.04538899287581444,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.6015,
      "step": 1161
    },
    {
      "epoch": 2.5779256794231835,
      "grad_norm": 0.03833644092082977,
      "learning_rate": 4.6419753086419746e-05,
      "loss": 0.5485,
      "step": 1162
    },
    {
      "epoch": 2.5801442041042706,
      "grad_norm": 0.0473506934940815,
      "learning_rate": 4.6172839506172835e-05,
      "loss": 0.7222,
      "step": 1163
    },
    {
      "epoch": 2.5823627287853577,
      "grad_norm": 0.06149815768003464,
      "learning_rate": 4.592592592592592e-05,
      "loss": 0.6284,
      "step": 1164
    },
    {
      "epoch": 2.5845812534664447,
      "grad_norm": 0.041904546320438385,
      "learning_rate": 4.567901234567901e-05,
      "loss": 0.4426,
      "step": 1165
    },
    {
      "epoch": 2.586799778147532,
      "grad_norm": 0.03167545422911644,
      "learning_rate": 4.543209876543209e-05,
      "loss": 0.4309,
      "step": 1166
    },
    {
      "epoch": 2.589018302828619,
      "grad_norm": 0.0546041764318943,
      "learning_rate": 4.518518518518519e-05,
      "loss": 0.7401,
      "step": 1167
    },
    {
      "epoch": 2.591236827509706,
      "grad_norm": 0.04976246878504753,
      "learning_rate": 4.493827160493827e-05,
      "loss": 0.5996,
      "step": 1168
    },
    {
      "epoch": 2.593455352190793,
      "grad_norm": 0.03541766107082367,
      "learning_rate": 4.469135802469135e-05,
      "loss": 0.4876,
      "step": 1169
    },
    {
      "epoch": 2.59567387687188,
      "grad_norm": 0.04107186943292618,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.5461,
      "step": 1170
    },
    {
      "epoch": 2.597892401552967,
      "grad_norm": 0.03863276168704033,
      "learning_rate": 4.419753086419753e-05,
      "loss": 0.5709,
      "step": 1171
    },
    {
      "epoch": 2.6001109262340543,
      "grad_norm": 0.042654141783714294,
      "learning_rate": 4.3950617283950617e-05,
      "loss": 0.6435,
      "step": 1172
    },
    {
      "epoch": 2.6023294509151413,
      "grad_norm": 0.05361958593130112,
      "learning_rate": 4.37037037037037e-05,
      "loss": 0.6117,
      "step": 1173
    },
    {
      "epoch": 2.6045479755962284,
      "grad_norm": 0.054624881595373154,
      "learning_rate": 4.3456790123456786e-05,
      "loss": 0.4655,
      "step": 1174
    },
    {
      "epoch": 2.6067665002773155,
      "grad_norm": 0.05888504534959793,
      "learning_rate": 4.3209876543209875e-05,
      "loss": 0.5505,
      "step": 1175
    },
    {
      "epoch": 2.6089850249584026,
      "grad_norm": 0.031220784410834312,
      "learning_rate": 4.296296296296296e-05,
      "loss": 0.5061,
      "step": 1176
    },
    {
      "epoch": 2.6112035496394896,
      "grad_norm": 0.08287351578474045,
      "learning_rate": 4.2716049382716045e-05,
      "loss": 0.6202,
      "step": 1177
    },
    {
      "epoch": 2.6134220743205767,
      "grad_norm": 0.04395187273621559,
      "learning_rate": 4.2469135802469126e-05,
      "loss": 0.6511,
      "step": 1178
    },
    {
      "epoch": 2.615640599001664,
      "grad_norm": 0.04738863557577133,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.6764,
      "step": 1179
    },
    {
      "epoch": 2.617859123682751,
      "grad_norm": 0.03939921408891678,
      "learning_rate": 4.19753086419753e-05,
      "loss": 0.5693,
      "step": 1180
    },
    {
      "epoch": 2.620077648363838,
      "grad_norm": 0.05549682676792145,
      "learning_rate": 4.172839506172839e-05,
      "loss": 0.5564,
      "step": 1181
    },
    {
      "epoch": 2.622296173044925,
      "grad_norm": 0.043845098465681076,
      "learning_rate": 4.148148148148147e-05,
      "loss": 0.6161,
      "step": 1182
    },
    {
      "epoch": 2.624514697726012,
      "grad_norm": 0.057637207210063934,
      "learning_rate": 4.123456790123457e-05,
      "loss": 0.3999,
      "step": 1183
    },
    {
      "epoch": 2.626733222407099,
      "grad_norm": 0.05239883065223694,
      "learning_rate": 4.098765432098765e-05,
      "loss": 0.605,
      "step": 1184
    },
    {
      "epoch": 2.6289517470881862,
      "grad_norm": 0.046402182430028915,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.6719,
      "step": 1185
    },
    {
      "epoch": 2.6311702717692733,
      "grad_norm": 0.043412696570158005,
      "learning_rate": 4.0493827160493827e-05,
      "loss": 0.4603,
      "step": 1186
    },
    {
      "epoch": 2.6333887964503604,
      "grad_norm": 0.06885641813278198,
      "learning_rate": 4.0246913580246915e-05,
      "loss": 0.7485,
      "step": 1187
    },
    {
      "epoch": 2.6356073211314475,
      "grad_norm": 0.0403292216360569,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.4452,
      "step": 1188
    },
    {
      "epoch": 2.6378258458125345,
      "grad_norm": 0.042746204882860184,
      "learning_rate": 3.975308641975308e-05,
      "loss": 0.5085,
      "step": 1189
    },
    {
      "epoch": 2.6400443704936216,
      "grad_norm": 0.032522063702344894,
      "learning_rate": 3.950617283950617e-05,
      "loss": 0.4674,
      "step": 1190
    },
    {
      "epoch": 2.6422628951747087,
      "grad_norm": 0.040723737329244614,
      "learning_rate": 3.9259259259259255e-05,
      "loss": 0.5206,
      "step": 1191
    },
    {
      "epoch": 2.6444814198557958,
      "grad_norm": 0.033490557223558426,
      "learning_rate": 3.901234567901234e-05,
      "loss": 0.5752,
      "step": 1192
    },
    {
      "epoch": 2.646699944536883,
      "grad_norm": 0.03831752762198448,
      "learning_rate": 3.8765432098765425e-05,
      "loss": 0.6909,
      "step": 1193
    },
    {
      "epoch": 2.64891846921797,
      "grad_norm": 0.05617349594831467,
      "learning_rate": 3.851851851851852e-05,
      "loss": 0.6123,
      "step": 1194
    },
    {
      "epoch": 2.651136993899057,
      "grad_norm": 0.0512884147465229,
      "learning_rate": 3.82716049382716e-05,
      "loss": 0.5268,
      "step": 1195
    },
    {
      "epoch": 2.653355518580144,
      "grad_norm": 0.06002720445394516,
      "learning_rate": 3.802469135802468e-05,
      "loss": 0.4804,
      "step": 1196
    },
    {
      "epoch": 2.655574043261231,
      "grad_norm": 0.06364443153142929,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.4915,
      "step": 1197
    },
    {
      "epoch": 2.657792567942318,
      "grad_norm": 0.041675858199596405,
      "learning_rate": 3.7530864197530867e-05,
      "loss": 0.5229,
      "step": 1198
    },
    {
      "epoch": 2.6600110926234053,
      "grad_norm": 0.04145793989300728,
      "learning_rate": 3.728395061728395e-05,
      "loss": 0.5997,
      "step": 1199
    },
    {
      "epoch": 2.6622296173044924,
      "grad_norm": 0.057132210582494736,
      "learning_rate": 3.703703703703703e-05,
      "loss": 0.693,
      "step": 1200
    },
    {
      "epoch": 2.6644481419855794,
      "grad_norm": 0.050540413707494736,
      "learning_rate": 3.679012345679012e-05,
      "loss": 0.5451,
      "step": 1201
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.03993694856762886,
      "learning_rate": 3.6543209876543207e-05,
      "loss": 0.6044,
      "step": 1202
    },
    {
      "epoch": 2.6688851913477536,
      "grad_norm": 0.04184935986995697,
      "learning_rate": 3.6296296296296295e-05,
      "loss": 0.7356,
      "step": 1203
    },
    {
      "epoch": 2.6711037160288407,
      "grad_norm": 0.039861008524894714,
      "learning_rate": 3.604938271604938e-05,
      "loss": 0.7951,
      "step": 1204
    },
    {
      "epoch": 2.6733222407099277,
      "grad_norm": 0.03638765588402748,
      "learning_rate": 3.5802469135802465e-05,
      "loss": 0.5195,
      "step": 1205
    },
    {
      "epoch": 2.675540765391015,
      "grad_norm": 0.0478421188890934,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.608,
      "step": 1206
    },
    {
      "epoch": 2.677759290072102,
      "grad_norm": 0.0816487967967987,
      "learning_rate": 3.5308641975308635e-05,
      "loss": 0.5288,
      "step": 1207
    },
    {
      "epoch": 2.679977814753189,
      "grad_norm": 0.05673297122120857,
      "learning_rate": 3.506172839506172e-05,
      "loss": 0.6614,
      "step": 1208
    },
    {
      "epoch": 2.682196339434276,
      "grad_norm": 0.05833018571138382,
      "learning_rate": 3.481481481481481e-05,
      "loss": 0.5429,
      "step": 1209
    },
    {
      "epoch": 2.684414864115363,
      "grad_norm": 0.04374930262565613,
      "learning_rate": 3.45679012345679e-05,
      "loss": 0.5273,
      "step": 1210
    },
    {
      "epoch": 2.68663338879645,
      "grad_norm": 0.03475876897573471,
      "learning_rate": 3.432098765432098e-05,
      "loss": 0.5639,
      "step": 1211
    },
    {
      "epoch": 2.6888519134775377,
      "grad_norm": 0.04394547641277313,
      "learning_rate": 3.407407407407407e-05,
      "loss": 0.6745,
      "step": 1212
    },
    {
      "epoch": 2.691070438158625,
      "grad_norm": 0.03394545987248421,
      "learning_rate": 3.382716049382716e-05,
      "loss": 0.5404,
      "step": 1213
    },
    {
      "epoch": 2.693288962839712,
      "grad_norm": 0.0445111021399498,
      "learning_rate": 3.3580246913580247e-05,
      "loss": 0.635,
      "step": 1214
    },
    {
      "epoch": 2.695507487520799,
      "grad_norm": 0.049827393144369125,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.545,
      "step": 1215
    },
    {
      "epoch": 2.697726012201886,
      "grad_norm": 0.050018250942230225,
      "learning_rate": 3.3086419753086417e-05,
      "loss": 0.6383,
      "step": 1216
    },
    {
      "epoch": 2.699944536882973,
      "grad_norm": 0.03852216526865959,
      "learning_rate": 3.2839506172839505e-05,
      "loss": 0.574,
      "step": 1217
    },
    {
      "epoch": 2.70216306156406,
      "grad_norm": 0.04717600718140602,
      "learning_rate": 3.2592592592592586e-05,
      "loss": 0.571,
      "step": 1218
    },
    {
      "epoch": 2.7043815862451472,
      "grad_norm": 0.04441976174712181,
      "learning_rate": 3.2345679012345675e-05,
      "loss": 0.6381,
      "step": 1219
    },
    {
      "epoch": 2.7066001109262343,
      "grad_norm": 0.0717201828956604,
      "learning_rate": 3.209876543209876e-05,
      "loss": 0.6707,
      "step": 1220
    },
    {
      "epoch": 2.7088186356073214,
      "grad_norm": 0.042869970202445984,
      "learning_rate": 3.185185185185185e-05,
      "loss": 0.5493,
      "step": 1221
    },
    {
      "epoch": 2.7110371602884085,
      "grad_norm": 0.04704178124666214,
      "learning_rate": 3.160493827160493e-05,
      "loss": 0.6525,
      "step": 1222
    },
    {
      "epoch": 2.7132556849694955,
      "grad_norm": 0.06668686121702194,
      "learning_rate": 3.135802469135802e-05,
      "loss": 0.6725,
      "step": 1223
    },
    {
      "epoch": 2.7154742096505826,
      "grad_norm": 0.04320479556918144,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.6942,
      "step": 1224
    },
    {
      "epoch": 2.7176927343316697,
      "grad_norm": 0.053923748433589935,
      "learning_rate": 3.08641975308642e-05,
      "loss": 0.6712,
      "step": 1225
    },
    {
      "epoch": 2.7199112590127568,
      "grad_norm": 0.03926434367895126,
      "learning_rate": 3.061728395061728e-05,
      "loss": 0.5531,
      "step": 1226
    },
    {
      "epoch": 2.722129783693844,
      "grad_norm": 0.04206327721476555,
      "learning_rate": 3.0370370370370368e-05,
      "loss": 0.6322,
      "step": 1227
    },
    {
      "epoch": 2.724348308374931,
      "grad_norm": 0.035735778510570526,
      "learning_rate": 3.0123456790123453e-05,
      "loss": 0.5936,
      "step": 1228
    },
    {
      "epoch": 2.726566833056018,
      "grad_norm": 0.04683126136660576,
      "learning_rate": 2.987654320987654e-05,
      "loss": 0.5849,
      "step": 1229
    },
    {
      "epoch": 2.728785357737105,
      "grad_norm": 0.04763888195157051,
      "learning_rate": 2.9629629629629627e-05,
      "loss": 0.498,
      "step": 1230
    },
    {
      "epoch": 2.731003882418192,
      "grad_norm": 0.04742740839719772,
      "learning_rate": 2.9382716049382715e-05,
      "loss": 0.6491,
      "step": 1231
    },
    {
      "epoch": 2.733222407099279,
      "grad_norm": 0.049570273607969284,
      "learning_rate": 2.9135802469135796e-05,
      "loss": 0.4266,
      "step": 1232
    },
    {
      "epoch": 2.7354409317803663,
      "grad_norm": 0.053373873233795166,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.5552,
      "step": 1233
    },
    {
      "epoch": 2.7376594564614534,
      "grad_norm": 0.036181122064590454,
      "learning_rate": 2.8641975308641973e-05,
      "loss": 0.5341,
      "step": 1234
    },
    {
      "epoch": 2.7398779811425404,
      "grad_norm": 0.047016702592372894,
      "learning_rate": 2.8395061728395058e-05,
      "loss": 0.4577,
      "step": 1235
    },
    {
      "epoch": 2.7420965058236275,
      "grad_norm": 0.034889232367277145,
      "learning_rate": 2.8148148148148147e-05,
      "loss": 0.5484,
      "step": 1236
    },
    {
      "epoch": 2.7443150305047146,
      "grad_norm": 0.04580210521817207,
      "learning_rate": 2.790123456790123e-05,
      "loss": 0.5942,
      "step": 1237
    },
    {
      "epoch": 2.7465335551858017,
      "grad_norm": 0.07487092167139053,
      "learning_rate": 2.765432098765432e-05,
      "loss": 0.6523,
      "step": 1238
    },
    {
      "epoch": 2.7487520798668887,
      "grad_norm": 0.08859751373529434,
      "learning_rate": 2.7407407407407405e-05,
      "loss": 0.5349,
      "step": 1239
    },
    {
      "epoch": 2.750970604547976,
      "grad_norm": 0.0426231287419796,
      "learning_rate": 2.7160493827160493e-05,
      "loss": 0.6897,
      "step": 1240
    },
    {
      "epoch": 2.753189129229063,
      "grad_norm": 0.040884021669626236,
      "learning_rate": 2.6913580246913578e-05,
      "loss": 0.6232,
      "step": 1241
    },
    {
      "epoch": 2.75540765391015,
      "grad_norm": 0.040149349719285965,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.5117,
      "step": 1242
    },
    {
      "epoch": 2.757626178591237,
      "grad_norm": 0.04139826074242592,
      "learning_rate": 2.6419753086419748e-05,
      "loss": 0.5414,
      "step": 1243
    },
    {
      "epoch": 2.759844703272324,
      "grad_norm": 0.04517665505409241,
      "learning_rate": 2.6172839506172837e-05,
      "loss": 0.6894,
      "step": 1244
    },
    {
      "epoch": 2.762063227953411,
      "grad_norm": 0.04626264050602913,
      "learning_rate": 2.592592592592592e-05,
      "loss": 0.6789,
      "step": 1245
    },
    {
      "epoch": 2.7642817526344983,
      "grad_norm": 0.038021594285964966,
      "learning_rate": 2.567901234567901e-05,
      "loss": 0.5559,
      "step": 1246
    },
    {
      "epoch": 2.7665002773155853,
      "grad_norm": 0.0544557124376297,
      "learning_rate": 2.5432098765432095e-05,
      "loss": 0.6317,
      "step": 1247
    },
    {
      "epoch": 2.7687188019966724,
      "grad_norm": 0.044251710176467896,
      "learning_rate": 2.5185185185185183e-05,
      "loss": 0.5095,
      "step": 1248
    },
    {
      "epoch": 2.7709373266777595,
      "grad_norm": 0.03485402837395668,
      "learning_rate": 2.4938271604938268e-05,
      "loss": 0.7032,
      "step": 1249
    },
    {
      "epoch": 2.7731558513588466,
      "grad_norm": 0.036658111959695816,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 0.5815,
      "step": 1250
    },
    {
      "epoch": 2.7753743760399336,
      "grad_norm": 0.045029882341623306,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.7131,
      "step": 1251
    },
    {
      "epoch": 2.7775929007210207,
      "grad_norm": 0.042537666857242584,
      "learning_rate": 2.419753086419753e-05,
      "loss": 0.5593,
      "step": 1252
    },
    {
      "epoch": 2.779811425402108,
      "grad_norm": 0.04805844649672508,
      "learning_rate": 2.3950617283950618e-05,
      "loss": 0.7943,
      "step": 1253
    },
    {
      "epoch": 2.782029950083195,
      "grad_norm": 0.049093712121248245,
      "learning_rate": 2.37037037037037e-05,
      "loss": 0.5271,
      "step": 1254
    },
    {
      "epoch": 2.784248474764282,
      "grad_norm": 0.030959073454141617,
      "learning_rate": 2.3456790123456788e-05,
      "loss": 0.551,
      "step": 1255
    },
    {
      "epoch": 2.786466999445369,
      "grad_norm": 0.04587647318840027,
      "learning_rate": 2.3209876543209873e-05,
      "loss": 0.6271,
      "step": 1256
    },
    {
      "epoch": 2.788685524126456,
      "grad_norm": 0.04436396434903145,
      "learning_rate": 2.296296296296296e-05,
      "loss": 0.7355,
      "step": 1257
    },
    {
      "epoch": 2.790904048807543,
      "grad_norm": 0.03463001549243927,
      "learning_rate": 2.2716049382716047e-05,
      "loss": 0.5476,
      "step": 1258
    },
    {
      "epoch": 2.7931225734886302,
      "grad_norm": 0.043275997042655945,
      "learning_rate": 2.2469135802469135e-05,
      "loss": 0.7732,
      "step": 1259
    },
    {
      "epoch": 2.7953410981697173,
      "grad_norm": 0.04601794108748436,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.6197,
      "step": 1260
    },
    {
      "epoch": 2.7975596228508044,
      "grad_norm": 0.06180325895547867,
      "learning_rate": 2.1975308641975308e-05,
      "loss": 0.4662,
      "step": 1261
    },
    {
      "epoch": 2.7997781475318915,
      "grad_norm": 0.057592809200286865,
      "learning_rate": 2.1728395061728393e-05,
      "loss": 0.6415,
      "step": 1262
    },
    {
      "epoch": 2.8019966722129785,
      "grad_norm": 0.04574737697839737,
      "learning_rate": 2.148148148148148e-05,
      "loss": 0.4613,
      "step": 1263
    },
    {
      "epoch": 2.8042151968940656,
      "grad_norm": 0.04672899469733238,
      "learning_rate": 2.1234567901234563e-05,
      "loss": 0.6369,
      "step": 1264
    },
    {
      "epoch": 2.8064337215751527,
      "grad_norm": 0.05607036501169205,
      "learning_rate": 2.098765432098765e-05,
      "loss": 0.7394,
      "step": 1265
    },
    {
      "epoch": 2.8086522462562398,
      "grad_norm": 0.0463699996471405,
      "learning_rate": 2.0740740740740737e-05,
      "loss": 0.547,
      "step": 1266
    },
    {
      "epoch": 2.810870770937327,
      "grad_norm": 0.041211433708667755,
      "learning_rate": 2.0493827160493825e-05,
      "loss": 0.7159,
      "step": 1267
    },
    {
      "epoch": 2.813089295618414,
      "grad_norm": 0.058875758200883865,
      "learning_rate": 2.0246913580246913e-05,
      "loss": 0.5315,
      "step": 1268
    },
    {
      "epoch": 2.815307820299501,
      "grad_norm": 0.038778290152549744,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.4877,
      "step": 1269
    },
    {
      "epoch": 2.817526344980588,
      "grad_norm": 0.04740240424871445,
      "learning_rate": 1.9753086419753087e-05,
      "loss": 0.5142,
      "step": 1270
    },
    {
      "epoch": 2.819744869661675,
      "grad_norm": 0.04323930665850639,
      "learning_rate": 1.950617283950617e-05,
      "loss": 0.5295,
      "step": 1271
    },
    {
      "epoch": 2.821963394342762,
      "grad_norm": 0.04316713660955429,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.5927,
      "step": 1272
    },
    {
      "epoch": 2.8241819190238493,
      "grad_norm": 0.040836069732904434,
      "learning_rate": 1.901234567901234e-05,
      "loss": 0.5633,
      "step": 1273
    },
    {
      "epoch": 2.8264004437049364,
      "grad_norm": 0.04833223670721054,
      "learning_rate": 1.8765432098765433e-05,
      "loss": 0.5935,
      "step": 1274
    },
    {
      "epoch": 2.8286189683860234,
      "grad_norm": 0.047674380242824554,
      "learning_rate": 1.8518518518518515e-05,
      "loss": 0.7315,
      "step": 1275
    },
    {
      "epoch": 2.8308374930671105,
      "grad_norm": 0.038580264896154404,
      "learning_rate": 1.8271604938271603e-05,
      "loss": 0.5386,
      "step": 1276
    },
    {
      "epoch": 2.8330560177481976,
      "grad_norm": 0.04538154602050781,
      "learning_rate": 1.802469135802469e-05,
      "loss": 0.6806,
      "step": 1277
    },
    {
      "epoch": 2.8352745424292847,
      "grad_norm": 0.04222889244556427,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.4843,
      "step": 1278
    },
    {
      "epoch": 2.8374930671103717,
      "grad_norm": 0.04233108088374138,
      "learning_rate": 1.753086419753086e-05,
      "loss": 0.5341,
      "step": 1279
    },
    {
      "epoch": 2.839711591791459,
      "grad_norm": 0.04549160599708557,
      "learning_rate": 1.728395061728395e-05,
      "loss": 0.5785,
      "step": 1280
    },
    {
      "epoch": 2.841930116472546,
      "grad_norm": 0.032927896827459335,
      "learning_rate": 1.7037037037037035e-05,
      "loss": 0.4491,
      "step": 1281
    },
    {
      "epoch": 2.844148641153633,
      "grad_norm": 0.04449811577796936,
      "learning_rate": 1.6790123456790123e-05,
      "loss": 0.7317,
      "step": 1282
    },
    {
      "epoch": 2.84636716583472,
      "grad_norm": 0.03668645769357681,
      "learning_rate": 1.6543209876543208e-05,
      "loss": 0.5154,
      "step": 1283
    },
    {
      "epoch": 2.848585690515807,
      "grad_norm": 0.04218946397304535,
      "learning_rate": 1.6296296296296293e-05,
      "loss": 0.4625,
      "step": 1284
    },
    {
      "epoch": 2.850804215196894,
      "grad_norm": 0.03924800828099251,
      "learning_rate": 1.604938271604938e-05,
      "loss": 0.58,
      "step": 1285
    },
    {
      "epoch": 2.8530227398779813,
      "grad_norm": 0.036413632333278656,
      "learning_rate": 1.5802469135802467e-05,
      "loss": 0.4978,
      "step": 1286
    },
    {
      "epoch": 2.8552412645590683,
      "grad_norm": 0.054489899426698685,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.4034,
      "step": 1287
    },
    {
      "epoch": 2.8574597892401554,
      "grad_norm": 0.036673806607723236,
      "learning_rate": 1.530864197530864e-05,
      "loss": 0.5359,
      "step": 1288
    },
    {
      "epoch": 2.8596783139212425,
      "grad_norm": 0.06210623309016228,
      "learning_rate": 1.5061728395061727e-05,
      "loss": 0.6413,
      "step": 1289
    },
    {
      "epoch": 2.8618968386023296,
      "grad_norm": 0.039670929312705994,
      "learning_rate": 1.4814814814814813e-05,
      "loss": 0.4899,
      "step": 1290
    },
    {
      "epoch": 2.8641153632834166,
      "grad_norm": 0.03689377009868622,
      "learning_rate": 1.4567901234567898e-05,
      "loss": 0.5401,
      "step": 1291
    },
    {
      "epoch": 2.8663338879645037,
      "grad_norm": 0.05128929018974304,
      "learning_rate": 1.4320987654320987e-05,
      "loss": 0.5293,
      "step": 1292
    },
    {
      "epoch": 2.868552412645591,
      "grad_norm": 0.0823260098695755,
      "learning_rate": 1.4074074074074073e-05,
      "loss": 0.7737,
      "step": 1293
    },
    {
      "epoch": 2.870770937326678,
      "grad_norm": 0.0485205352306366,
      "learning_rate": 1.382716049382716e-05,
      "loss": 0.6429,
      "step": 1294
    },
    {
      "epoch": 2.872989462007765,
      "grad_norm": 0.043115995824337006,
      "learning_rate": 1.3580246913580247e-05,
      "loss": 0.6624,
      "step": 1295
    },
    {
      "epoch": 2.875207986688852,
      "grad_norm": 0.0446871817111969,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.5427,
      "step": 1296
    },
    {
      "epoch": 2.877426511369939,
      "grad_norm": 0.042325280606746674,
      "learning_rate": 1.3086419753086418e-05,
      "loss": 0.5383,
      "step": 1297
    },
    {
      "epoch": 2.879645036051026,
      "grad_norm": 0.046466726809740067,
      "learning_rate": 1.2839506172839505e-05,
      "loss": 0.6947,
      "step": 1298
    },
    {
      "epoch": 2.8818635607321132,
      "grad_norm": 0.059913478791713715,
      "learning_rate": 1.2592592592592592e-05,
      "loss": 0.7451,
      "step": 1299
    },
    {
      "epoch": 2.8840820854132003,
      "grad_norm": 0.04625663161277771,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 0.5347,
      "step": 1300
    },
    {
      "epoch": 2.8863006100942874,
      "grad_norm": 0.05530298128724098,
      "learning_rate": 1.2098765432098765e-05,
      "loss": 0.6392,
      "step": 1301
    },
    {
      "epoch": 2.8885191347753745,
      "grad_norm": 0.047867126762866974,
      "learning_rate": 1.185185185185185e-05,
      "loss": 0.493,
      "step": 1302
    },
    {
      "epoch": 2.8907376594564616,
      "grad_norm": 0.07963269203901291,
      "learning_rate": 1.1604938271604937e-05,
      "loss": 0.6336,
      "step": 1303
    },
    {
      "epoch": 2.8929561841375486,
      "grad_norm": 0.042634591460227966,
      "learning_rate": 1.1358024691358023e-05,
      "loss": 0.5981,
      "step": 1304
    },
    {
      "epoch": 2.8951747088186357,
      "grad_norm": 0.07128524780273438,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.542,
      "step": 1305
    },
    {
      "epoch": 2.8973932334997228,
      "grad_norm": 0.03820500895380974,
      "learning_rate": 1.0864197530864197e-05,
      "loss": 0.5731,
      "step": 1306
    },
    {
      "epoch": 2.89961175818081,
      "grad_norm": 0.049363672733306885,
      "learning_rate": 1.0617283950617282e-05,
      "loss": 0.6045,
      "step": 1307
    },
    {
      "epoch": 2.901830282861897,
      "grad_norm": 0.052022259682416916,
      "learning_rate": 1.0370370370370368e-05,
      "loss": 0.6209,
      "step": 1308
    },
    {
      "epoch": 2.904048807542984,
      "grad_norm": 0.04577159136533737,
      "learning_rate": 1.0123456790123457e-05,
      "loss": 0.5581,
      "step": 1309
    },
    {
      "epoch": 2.906267332224071,
      "grad_norm": 0.03305206820368767,
      "learning_rate": 9.876543209876543e-06,
      "loss": 0.3672,
      "step": 1310
    },
    {
      "epoch": 2.908485856905158,
      "grad_norm": 0.032856106758117676,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.5863,
      "step": 1311
    },
    {
      "epoch": 2.9107043815862452,
      "grad_norm": 0.04961647838354111,
      "learning_rate": 9.382716049382717e-06,
      "loss": 0.6017,
      "step": 1312
    },
    {
      "epoch": 2.9129229062673323,
      "grad_norm": 0.04911202937364578,
      "learning_rate": 9.135802469135802e-06,
      "loss": 0.5172,
      "step": 1313
    },
    {
      "epoch": 2.9151414309484194,
      "grad_norm": 0.03537707030773163,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.551,
      "step": 1314
    },
    {
      "epoch": 2.9173599556295065,
      "grad_norm": 0.07286648452281952,
      "learning_rate": 8.641975308641975e-06,
      "loss": 0.565,
      "step": 1315
    },
    {
      "epoch": 2.9195784803105935,
      "grad_norm": 0.03762354329228401,
      "learning_rate": 8.395061728395062e-06,
      "loss": 0.8136,
      "step": 1316
    },
    {
      "epoch": 2.9217970049916806,
      "grad_norm": 0.04370323196053505,
      "learning_rate": 8.148148148148147e-06,
      "loss": 0.644,
      "step": 1317
    },
    {
      "epoch": 2.9240155296727677,
      "grad_norm": 0.03825458139181137,
      "learning_rate": 7.901234567901233e-06,
      "loss": 0.5718,
      "step": 1318
    },
    {
      "epoch": 2.9262340543538548,
      "grad_norm": 0.03614470362663269,
      "learning_rate": 7.65432098765432e-06,
      "loss": 0.5574,
      "step": 1319
    },
    {
      "epoch": 2.928452579034942,
      "grad_norm": 0.05544061213731766,
      "learning_rate": 7.407407407407407e-06,
      "loss": 0.6487,
      "step": 1320
    },
    {
      "epoch": 2.930671103716029,
      "grad_norm": 0.055062491446733475,
      "learning_rate": 7.160493827160493e-06,
      "loss": 0.7268,
      "step": 1321
    },
    {
      "epoch": 2.932889628397116,
      "grad_norm": 0.039739031344652176,
      "learning_rate": 6.91358024691358e-06,
      "loss": 0.5391,
      "step": 1322
    },
    {
      "epoch": 2.935108153078203,
      "grad_norm": 0.04464324563741684,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5606,
      "step": 1323
    },
    {
      "epoch": 2.93732667775929,
      "grad_norm": 0.04252636060118675,
      "learning_rate": 6.4197530864197525e-06,
      "loss": 0.5389,
      "step": 1324
    },
    {
      "epoch": 2.939545202440377,
      "grad_norm": 0.048218585550785065,
      "learning_rate": 6.172839506172839e-06,
      "loss": 0.6824,
      "step": 1325
    },
    {
      "epoch": 2.9417637271214643,
      "grad_norm": 0.039460841566324234,
      "learning_rate": 5.925925925925925e-06,
      "loss": 0.591,
      "step": 1326
    },
    {
      "epoch": 2.9439822518025514,
      "grad_norm": 0.047840118408203125,
      "learning_rate": 5.679012345679012e-06,
      "loss": 0.6378,
      "step": 1327
    },
    {
      "epoch": 2.9462007764836384,
      "grad_norm": 0.04324064403772354,
      "learning_rate": 5.432098765432098e-06,
      "loss": 0.693,
      "step": 1328
    },
    {
      "epoch": 2.9484193011647255,
      "grad_norm": 0.05165842920541763,
      "learning_rate": 5.185185185185184e-06,
      "loss": 0.4814,
      "step": 1329
    },
    {
      "epoch": 2.9506378258458126,
      "grad_norm": 0.04023590683937073,
      "learning_rate": 4.938271604938272e-06,
      "loss": 0.6161,
      "step": 1330
    },
    {
      "epoch": 2.9528563505268997,
      "grad_norm": 0.060393720865249634,
      "learning_rate": 4.691358024691358e-06,
      "loss": 0.6807,
      "step": 1331
    },
    {
      "epoch": 2.9550748752079867,
      "grad_norm": 0.034441426396369934,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.577,
      "step": 1332
    },
    {
      "epoch": 2.957293399889074,
      "grad_norm": 0.060275763273239136,
      "learning_rate": 4.197530864197531e-06,
      "loss": 0.5678,
      "step": 1333
    },
    {
      "epoch": 2.959511924570161,
      "grad_norm": 0.046736739575862885,
      "learning_rate": 3.950617283950617e-06,
      "loss": 0.619,
      "step": 1334
    },
    {
      "epoch": 2.961730449251248,
      "grad_norm": 0.038540858775377274,
      "learning_rate": 3.7037037037037033e-06,
      "loss": 0.6857,
      "step": 1335
    },
    {
      "epoch": 2.963948973932335,
      "grad_norm": 0.049309924244880676,
      "learning_rate": 3.45679012345679e-06,
      "loss": 0.7506,
      "step": 1336
    },
    {
      "epoch": 2.966167498613422,
      "grad_norm": 0.0493754968047142,
      "learning_rate": 3.2098765432098762e-06,
      "loss": 0.5518,
      "step": 1337
    },
    {
      "epoch": 2.968386023294509,
      "grad_norm": 0.03662782534956932,
      "learning_rate": 2.9629629629629625e-06,
      "loss": 0.5282,
      "step": 1338
    },
    {
      "epoch": 2.9706045479755963,
      "grad_norm": 0.037066757678985596,
      "learning_rate": 2.716049382716049e-06,
      "loss": 0.4653,
      "step": 1339
    },
    {
      "epoch": 2.9728230726566833,
      "grad_norm": 0.04098521173000336,
      "learning_rate": 2.469135802469136e-06,
      "loss": 0.6656,
      "step": 1340
    },
    {
      "epoch": 2.9750415973377704,
      "grad_norm": 0.04167475551366806,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.6079,
      "step": 1341
    },
    {
      "epoch": 2.9772601220188575,
      "grad_norm": 0.0731402263045311,
      "learning_rate": 1.9753086419753083e-06,
      "loss": 0.7023,
      "step": 1342
    },
    {
      "epoch": 2.9794786466999446,
      "grad_norm": 0.05055250972509384,
      "learning_rate": 1.728395061728395e-06,
      "loss": 0.5526,
      "step": 1343
    },
    {
      "epoch": 2.9816971713810316,
      "grad_norm": 0.06800078600645065,
      "learning_rate": 1.4814814814814812e-06,
      "loss": 0.6206,
      "step": 1344
    },
    {
      "epoch": 2.9839156960621187,
      "grad_norm": 0.03610071912407875,
      "learning_rate": 1.234567901234568e-06,
      "loss": 0.7568,
      "step": 1345
    },
    {
      "epoch": 2.986134220743206,
      "grad_norm": 0.07225791364908218,
      "learning_rate": 9.876543209876542e-07,
      "loss": 0.7953,
      "step": 1346
    },
    {
      "epoch": 2.988352745424293,
      "grad_norm": 0.03866912052035332,
      "learning_rate": 7.407407407407406e-07,
      "loss": 0.5238,
      "step": 1347
    },
    {
      "epoch": 2.99057127010538,
      "grad_norm": 0.0409688875079155,
      "learning_rate": 4.938271604938271e-07,
      "loss": 0.5274,
      "step": 1348
    },
    {
      "epoch": 2.992789794786467,
      "grad_norm": 0.051803864538669586,
      "learning_rate": 2.4691358024691354e-07,
      "loss": 0.5913,
      "step": 1349
    },
    {
      "epoch": 2.995008319467554,
      "grad_norm": 0.03417398780584335,
      "learning_rate": 0.0,
      "loss": 0.4938,
      "step": 1350
    },
    {
      "epoch": 2.995008319467554,
      "step": 1350,
      "total_flos": 5.0830773065036595e+17,
      "train_loss": 0.6010428385822861,
      "train_runtime": 4607.4026,
      "train_samples_per_second": 4.695,
      "train_steps_per_second": 0.293
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1350,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.0830773065036595e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}