{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.5034168564920274,
  "eval_steps": 500,
  "global_step": 660,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002277904328018223,
      "grad_norm": 0.0,
      "learning_rate": 1.999771975829438e-05,
      "loss": 1.9165,
      "step": 1
    },
    {
      "epoch": 0.004555808656036446,
      "grad_norm": 0.0,
      "learning_rate": 1.9995434832230086e-05,
      "loss": 2.8215,
      "step": 2
    },
    {
      "epoch": 0.00683371298405467,
      "grad_norm": 0.0,
      "learning_rate": 1.9993145207357477e-05,
      "loss": 2.8218,
      "step": 3
    },
    {
      "epoch": 0.009111617312072893,
      "grad_norm": 0.0,
      "learning_rate": 1.999085086916743e-05,
      "loss": 2.25,
      "step": 4
    },
    {
      "epoch": 0.011389521640091117,
      "grad_norm": 0.0,
      "learning_rate": 1.9988551803091016e-05,
      "loss": 2.0216,
      "step": 5
    },
    {
      "epoch": 0.01366742596810934,
      "grad_norm": 0.0,
      "learning_rate": 1.99862479944992e-05,
      "loss": 2.0384,
      "step": 6
    },
    {
      "epoch": 0.015945330296127564,
      "grad_norm": 0.0,
      "learning_rate": 1.9983939428702538e-05,
      "loss": 2.0138,
      "step": 7
    },
    {
      "epoch": 0.018223234624145785,
      "grad_norm": 0.0,
      "learning_rate": 1.9981626090950854e-05,
      "loss": 2.1282,
      "step": 8
    },
    {
      "epoch": 0.02050113895216401,
      "grad_norm": 0.0,
      "learning_rate": 1.9979307966432923e-05,
      "loss": 2.1009,
      "step": 9
    },
    {
      "epoch": 0.022779043280182234,
      "grad_norm": 0.0,
      "learning_rate": 1.9976985040276184e-05,
      "loss": 1.9757,
      "step": 10
    },
    {
      "epoch": 0.025056947608200455,
      "grad_norm": 0.0,
      "learning_rate": 1.9974657297546367e-05,
      "loss": 1.9489,
      "step": 11
    },
    {
      "epoch": 0.02733485193621868,
      "grad_norm": 0.0,
      "learning_rate": 1.9972324723247235e-05,
      "loss": 2.0921,
      "step": 12
    },
    {
      "epoch": 0.029612756264236904,
      "grad_norm": 0.0,
      "learning_rate": 1.9969987302320216e-05,
      "loss": 2.0912,
      "step": 13
    },
    {
      "epoch": 0.03189066059225513,
      "grad_norm": 0.0,
      "learning_rate": 1.9967645019644098e-05,
      "loss": 1.979,
      "step": 14
    },
    {
      "epoch": 0.03416856492027335,
      "grad_norm": 0.0,
      "learning_rate": 1.99652978600347e-05,
      "loss": 1.9276,
      "step": 15
    },
    {
      "epoch": 0.03644646924829157,
      "grad_norm": 0.0,
      "learning_rate": 1.996294580824456e-05,
      "loss": 2.0417,
      "step": 16
    },
    {
      "epoch": 0.0387243735763098,
      "grad_norm": 0.0,
      "learning_rate": 1.996058884896256e-05,
      "loss": 1.9939,
      "step": 17
    },
    {
      "epoch": 0.04100227790432802,
      "grad_norm": 0.0,
      "learning_rate": 1.9958226966813648e-05,
      "loss": 2.0766,
      "step": 18
    },
    {
      "epoch": 0.04328018223234624,
      "grad_norm": 0.0,
      "learning_rate": 1.9955860146358464e-05,
      "loss": 1.9781,
      "step": 19
    },
    {
      "epoch": 0.04555808656036447,
      "grad_norm": 0.0,
      "learning_rate": 1.9953488372093024e-05,
      "loss": 2.0005,
      "step": 20
    },
    {
      "epoch": 0.04783599088838269,
      "grad_norm": 0.0,
      "learning_rate": 1.995111162844838e-05,
      "loss": 2.006,
      "step": 21
    },
    {
      "epoch": 0.05011389521640091,
      "grad_norm": 0.0,
      "learning_rate": 1.994872989979026e-05,
      "loss": 1.9772,
      "step": 22
    },
    {
      "epoch": 0.05239179954441914,
      "grad_norm": 0.0,
      "learning_rate": 1.9946343170418757e-05,
      "loss": 2.0346,
      "step": 23
    },
    {
      "epoch": 0.05466970387243736,
      "grad_norm": 0.0,
      "learning_rate": 1.9943951424567963e-05,
      "loss": 1.8697,
      "step": 24
    },
    {
      "epoch": 0.05694760820045558,
      "grad_norm": 0.0,
      "learning_rate": 1.994155464640561e-05,
      "loss": 2.0224,
      "step": 25
    },
    {
      "epoch": 0.05922551252847381,
      "grad_norm": 0.0,
      "learning_rate": 1.9939152820032765e-05,
      "loss": 1.9172,
      "step": 26
    },
    {
      "epoch": 0.06150341685649203,
      "grad_norm": 0.0,
      "learning_rate": 1.9936745929483427e-05,
      "loss": 1.948,
      "step": 27
    },
    {
      "epoch": 0.06378132118451026,
      "grad_norm": 0.0,
      "learning_rate": 1.9934333958724203e-05,
      "loss": 1.9458,
      "step": 28
    },
    {
      "epoch": 0.06605922551252848,
      "grad_norm": 0.0,
      "learning_rate": 1.993191689165395e-05,
      "loss": 1.8594,
      "step": 29
    },
    {
      "epoch": 0.0683371298405467,
      "grad_norm": 0.0,
      "learning_rate": 1.992949471210341e-05,
      "loss": 1.9468,
      "step": 30
    },
    {
      "epoch": 0.07061503416856492,
      "grad_norm": 0.0,
      "learning_rate": 1.9927067403834844e-05,
      "loss": 1.9019,
      "step": 31
    },
    {
      "epoch": 0.07289293849658314,
      "grad_norm": 0.0,
      "learning_rate": 1.9924634950541687e-05,
      "loss": 1.9821,
      "step": 32
    },
    {
      "epoch": 0.07517084282460136,
      "grad_norm": 0.0,
      "learning_rate": 1.992219733584817e-05,
      "loss": 1.8991,
      "step": 33
    },
    {
      "epoch": 0.0774487471526196,
      "grad_norm": 0.0,
      "learning_rate": 1.9919754543308948e-05,
      "loss": 1.97,
      "step": 34
    },
    {
      "epoch": 0.07972665148063782,
      "grad_norm": 0.0,
      "learning_rate": 1.9917306556408743e-05,
      "loss": 1.925,
      "step": 35
    },
    {
      "epoch": 0.08200455580865604,
      "grad_norm": 0.0,
      "learning_rate": 1.991485335856197e-05,
      "loss": 1.8711,
      "step": 36
    },
    {
      "epoch": 0.08428246013667426,
      "grad_norm": 0.0,
      "learning_rate": 1.991239493311235e-05,
      "loss": 1.8825,
      "step": 37
    },
    {
      "epoch": 0.08656036446469248,
      "grad_norm": 0.0,
      "learning_rate": 1.9909931263332544e-05,
      "loss": 1.8301,
      "step": 38
    },
    {
      "epoch": 0.0888382687927107,
      "grad_norm": 0.0,
      "learning_rate": 1.9907462332423777e-05,
      "loss": 1.8035,
      "step": 39
    },
    {
      "epoch": 0.09111617312072894,
      "grad_norm": 0.0,
      "learning_rate": 1.990498812351544e-05,
      "loss": 1.8363,
      "step": 40
    },
    {
      "epoch": 0.09339407744874716,
      "grad_norm": 0.0,
      "learning_rate": 1.9902508619664727e-05,
      "loss": 1.9067,
      "step": 41
    },
    {
      "epoch": 0.09567198177676538,
      "grad_norm": 0.0,
      "learning_rate": 1.9900023803856224e-05,
      "loss": 1.921,
      "step": 42
    },
    {
      "epoch": 0.0979498861047836,
      "grad_norm": 0.0,
      "learning_rate": 1.9897533659001547e-05,
      "loss": 1.862,
      "step": 43
    },
    {
      "epoch": 0.10022779043280182,
      "grad_norm": 0.0,
      "learning_rate": 1.9895038167938934e-05,
      "loss": 1.9782,
      "step": 44
    },
    {
      "epoch": 0.10250569476082004,
      "grad_norm": 0.0,
      "learning_rate": 1.9892537313432837e-05,
      "loss": 2.0042,
      "step": 45
    },
    {
      "epoch": 0.10478359908883828,
      "grad_norm": 0.0,
      "learning_rate": 1.989003107817356e-05,
      "loss": 1.9677,
      "step": 46
    },
    {
      "epoch": 0.1070615034168565,
      "grad_norm": 0.0,
      "learning_rate": 1.9887519444776835e-05,
      "loss": 1.7879,
      "step": 47
    },
    {
      "epoch": 0.10933940774487472,
      "grad_norm": 0.0,
      "learning_rate": 1.988500239578342e-05,
      "loss": 1.9285,
      "step": 48
    },
    {
      "epoch": 0.11161731207289294,
      "grad_norm": 0.0,
      "learning_rate": 1.9882479913658713e-05,
      "loss": 1.8736,
      "step": 49
    },
    {
      "epoch": 0.11389521640091116,
      "grad_norm": 0.0,
      "learning_rate": 1.9879951980792316e-05,
      "loss": 1.9533,
      "step": 50
    },
    {
      "epoch": 0.11617312072892938,
      "grad_norm": 0.0,
      "learning_rate": 1.987741857949766e-05,
      "loss": 1.8078,
      "step": 51
    },
    {
      "epoch": 0.11845102505694761,
      "grad_norm": 0.0,
      "learning_rate": 1.9874879692011553e-05,
      "loss": 1.9007,
      "step": 52
    },
    {
      "epoch": 0.12072892938496584,
      "grad_norm": 0.0,
      "learning_rate": 1.98723353004938e-05,
      "loss": 1.9432,
      "step": 53
    },
    {
      "epoch": 0.12300683371298406,
      "grad_norm": 0.0,
      "learning_rate": 1.9869785387026768e-05,
      "loss": 1.9253,
      "step": 54
    },
    {
      "epoch": 0.1252847380410023,
      "grad_norm": 0.0,
      "learning_rate": 1.986722993361497e-05,
      "loss": 1.723,
      "step": 55
    },
    {
      "epoch": 0.1275626423690205,
      "grad_norm": 0.0,
      "learning_rate": 1.986466892218463e-05,
      "loss": 1.8857,
      "step": 56
    },
    {
      "epoch": 0.12984054669703873,
      "grad_norm": 0.0,
      "learning_rate": 1.9862102334583286e-05,
      "loss": 1.828,
      "step": 57
    },
    {
      "epoch": 0.13211845102505695,
      "grad_norm": 0.0,
      "learning_rate": 1.9859530152579322e-05,
      "loss": 2.0451,
      "step": 58
    },
    {
      "epoch": 0.13439635535307518,
      "grad_norm": 0.0,
      "learning_rate": 1.985695235786156e-05,
      "loss": 1.9004,
      "step": 59
    },
    {
      "epoch": 0.1366742596810934,
      "grad_norm": 0.0,
      "learning_rate": 1.9854368932038836e-05,
      "loss": 1.8175,
      "step": 60
    },
    {
      "epoch": 0.13895216400911162,
      "grad_norm": 0.0,
      "learning_rate": 1.9851779856639535e-05,
      "loss": 1.8686,
      "step": 61
    },
    {
      "epoch": 0.14123006833712984,
      "grad_norm": 0.0,
      "learning_rate": 1.9849185113111166e-05,
      "loss": 1.812,
      "step": 62
    },
    {
      "epoch": 0.14350797266514806,
      "grad_norm": 0.0,
      "learning_rate": 1.984658468281992e-05,
      "loss": 1.9148,
      "step": 63
    },
    {
      "epoch": 0.14578587699316628,
      "grad_norm": 0.0,
      "learning_rate": 1.9843978547050223e-05,
      "loss": 1.9053,
      "step": 64
    },
    {
      "epoch": 0.1480637813211845,
      "grad_norm": 0.0,
      "learning_rate": 1.9841366687004273e-05,
      "loss": 1.8183,
      "step": 65
    },
    {
      "epoch": 0.15034168564920272,
      "grad_norm": 0.0,
      "learning_rate": 1.9838749083801612e-05,
      "loss": 1.8301,
      "step": 66
    },
    {
      "epoch": 0.15261958997722094,
      "grad_norm": 0.0,
      "learning_rate": 1.9836125718478662e-05,
      "loss": 1.8372,
      "step": 67
    },
    {
      "epoch": 0.1548974943052392,
      "grad_norm": 0.0,
      "learning_rate": 1.983349657198825e-05,
      "loss": 2.0162,
      "step": 68
    },
    {
      "epoch": 0.1571753986332574,
      "grad_norm": 0.0,
      "learning_rate": 1.9830861625199166e-05,
      "loss": 1.961,
      "step": 69
    },
    {
      "epoch": 0.15945330296127563,
      "grad_norm": 0.0,
      "learning_rate": 1.9828220858895707e-05,
      "loss": 1.8757,
      "step": 70
    },
    {
      "epoch": 0.16173120728929385,
      "grad_norm": 0.0,
      "learning_rate": 1.982557425377718e-05,
      "loss": 1.731,
      "step": 71
    },
    {
      "epoch": 0.16400911161731208,
      "grad_norm": 0.0,
      "learning_rate": 1.9822921790457455e-05,
      "loss": 2.0243,
      "step": 72
    },
    {
      "epoch": 0.1662870159453303,
      "grad_norm": 0.0,
      "learning_rate": 1.9820263449464484e-05,
      "loss": 1.952,
      "step": 73
    },
    {
      "epoch": 0.16856492027334852,
      "grad_norm": 0.0,
      "learning_rate": 1.9817599211239834e-05,
      "loss": 1.8578,
      "step": 74
    },
    {
      "epoch": 0.17084282460136674,
      "grad_norm": 0.0,
      "learning_rate": 1.9814929056138188e-05,
      "loss": 1.875,
      "step": 75
    },
    {
      "epoch": 0.17312072892938496,
      "grad_norm": 0.0,
      "learning_rate": 1.9812252964426878e-05,
      "loss": 1.9307,
      "step": 76
    },
    {
      "epoch": 0.17539863325740318,
      "grad_norm": 0.0,
      "learning_rate": 1.9809570916285398e-05,
      "loss": 1.9268,
      "step": 77
    },
    {
      "epoch": 0.1776765375854214,
      "grad_norm": 0.0,
      "learning_rate": 1.9806882891804902e-05,
      "loss": 1.868,
      "step": 78
    },
    {
      "epoch": 0.17995444191343962,
      "grad_norm": 0.0,
      "learning_rate": 1.9804188870987733e-05,
      "loss": 2.0597,
      "step": 79
    },
    {
      "epoch": 0.18223234624145787,
      "grad_norm": 0.0,
      "learning_rate": 1.98014888337469e-05,
      "loss": 1.9474,
      "step": 80
    },
    {
      "epoch": 0.1845102505694761,
      "grad_norm": 0.0,
      "learning_rate": 1.9798782759905605e-05,
      "loss": 1.8404,
      "step": 81
    },
    {
      "epoch": 0.1867881548974943,
      "grad_norm": 0.0,
      "learning_rate": 1.9796070629196717e-05,
      "loss": 1.7861,
      "step": 82
    },
    {
      "epoch": 0.18906605922551253,
      "grad_norm": 0.0,
      "learning_rate": 1.9793352421262293e-05,
      "loss": 1.8087,
      "step": 83
    },
    {
      "epoch": 0.19134396355353075,
      "grad_norm": 0.0,
      "learning_rate": 1.9790628115653045e-05,
      "loss": 1.9031,
      "step": 84
    },
    {
      "epoch": 0.19362186788154898,
      "grad_norm": 0.0,
      "learning_rate": 1.9787897691827823e-05,
      "loss": 1.9287,
      "step": 85
    },
    {
      "epoch": 0.1958997722095672,
      "grad_norm": 0.0,
      "learning_rate": 1.9785161129153136e-05,
      "loss": 1.9303,
      "step": 86
    },
    {
      "epoch": 0.19817767653758542,
      "grad_norm": 0.0,
      "learning_rate": 1.978241840690259e-05,
      "loss": 1.7859,
      "step": 87
    },
    {
      "epoch": 0.20045558086560364,
      "grad_norm": 0.0,
      "learning_rate": 1.9779669504256388e-05,
      "loss": 1.7389,
      "step": 88
    },
    {
      "epoch": 0.20273348519362186,
      "grad_norm": 0.0,
      "learning_rate": 1.977691440030079e-05,
      "loss": 1.9208,
      "step": 89
    },
    {
      "epoch": 0.20501138952164008,
      "grad_norm": 0.0,
      "learning_rate": 1.9774153074027606e-05,
      "loss": 1.9071,
      "step": 90
    },
    {
      "epoch": 0.2072892938496583,
      "grad_norm": 0.0,
      "learning_rate": 1.977138550433363e-05,
      "loss": 1.8619,
      "step": 91
    },
    {
      "epoch": 0.20956719817767655,
      "grad_norm": 0.0,
      "learning_rate": 1.9768611670020123e-05,
      "loss": 1.6974,
      "step": 92
    },
    {
      "epoch": 0.21184510250569477,
      "grad_norm": 0.0,
      "learning_rate": 1.9765831549792272e-05,
      "loss": 1.8706,
      "step": 93
    },
    {
      "epoch": 0.214123006833713,
      "grad_norm": 0.0,
      "learning_rate": 1.9763045122258635e-05,
      "loss": 1.7,
      "step": 94
    },
    {
      "epoch": 0.2164009111617312,
      "grad_norm": 0.0,
      "learning_rate": 1.97602523659306e-05,
      "loss": 1.7257,
      "step": 95
    },
    {
      "epoch": 0.21867881548974943,
      "grad_norm": 0.0,
      "learning_rate": 1.975745325922183e-05,
      "loss": 1.8205,
      "step": 96
    },
    {
      "epoch": 0.22095671981776766,
      "grad_norm": 0.0,
      "learning_rate": 1.9754647780447705e-05,
      "loss": 1.9095,
      "step": 97
    },
    {
      "epoch": 0.22323462414578588,
      "grad_norm": 0.0,
      "learning_rate": 1.9751835907824768e-05,
      "loss": 1.8607,
      "step": 98
    },
    {
      "epoch": 0.2255125284738041,
      "grad_norm": 0.0,
      "learning_rate": 1.974901761947015e-05,
      "loss": 1.8098,
      "step": 99
    },
    {
      "epoch": 0.22779043280182232,
      "grad_norm": 0.0,
      "learning_rate": 1.9746192893401017e-05,
      "loss": 1.816,
      "step": 100
    },
    {
      "epoch": 0.23006833712984054,
      "grad_norm": 0.0,
      "learning_rate": 1.9743361707533988e-05,
      "loss": 1.8485,
      "step": 101
    },
    {
      "epoch": 0.23234624145785876,
      "grad_norm": 0.0,
      "learning_rate": 1.9740524039684558e-05,
      "loss": 1.6743,
      "step": 102
    },
    {
      "epoch": 0.23462414578587698,
      "grad_norm": 0.0,
      "learning_rate": 1.9737679867566535e-05,
      "loss": 1.9887,
      "step": 103
    },
    {
      "epoch": 0.23690205011389523,
      "grad_norm": 0.0,
      "learning_rate": 1.9734829168791436e-05,
      "loss": 1.8318,
      "step": 104
    },
    {
      "epoch": 0.23917995444191345,
      "grad_norm": 0.0,
      "learning_rate": 1.9731971920867902e-05,
      "loss": 1.9805,
      "step": 105
    },
    {
      "epoch": 0.24145785876993167,
      "grad_norm": 0.0,
      "learning_rate": 1.9729108101201127e-05,
      "loss": 1.7662,
      "step": 106
    },
    {
      "epoch": 0.2437357630979499,
      "grad_norm": 0.0,
      "learning_rate": 1.9726237687092234e-05,
      "loss": 1.9131,
      "step": 107
    },
    {
      "epoch": 0.2460136674259681,
      "grad_norm": 0.0,
      "learning_rate": 1.9723360655737707e-05,
      "loss": 1.7177,
      "step": 108
    },
    {
      "epoch": 0.24829157175398633,
      "grad_norm": 0.0,
      "learning_rate": 1.972047698422875e-05,
      "loss": 1.8605,
      "step": 109
    },
    {
      "epoch": 0.2505694760820046,
      "grad_norm": 0.0,
      "learning_rate": 1.9717586649550706e-05,
      "loss": 1.9709,
      "step": 110
    },
    {
      "epoch": 0.2528473804100228,
      "grad_norm": 0.0,
      "learning_rate": 1.9714689628582445e-05,
      "loss": 1.7844,
      "step": 111
    },
    {
      "epoch": 0.255125284738041,
      "grad_norm": 0.0,
      "learning_rate": 1.971178589809573e-05,
      "loss": 1.8793,
      "step": 112
    },
    {
      "epoch": 0.25740318906605925,
      "grad_norm": 0.0,
      "learning_rate": 1.9708875434754607e-05,
      "loss": 1.993,
      "step": 113
    },
    {
      "epoch": 0.25968109339407747,
      "grad_norm": 0.0,
      "learning_rate": 1.9705958215114784e-05,
      "loss": 1.8287,
      "step": 114
    },
    {
      "epoch": 0.2619589977220957,
      "grad_norm": 0.0,
      "learning_rate": 1.9703034215622984e-05,
      "loss": 1.7294,
      "step": 115
    },
    {
      "epoch": 0.2642369020501139,
      "grad_norm": 0.0,
      "learning_rate": 1.970010341261634e-05,
      "loss": 1.7535,
      "step": 116
    },
    {
      "epoch": 0.26651480637813213,
      "grad_norm": 0.0,
      "learning_rate": 1.969716578232173e-05,
      "loss": 1.8323,
      "step": 117
    },
    {
      "epoch": 0.26879271070615035,
      "grad_norm": 0.0,
      "learning_rate": 1.969422130085515e-05,
      "loss": 1.8709,
      "step": 118
    },
    {
      "epoch": 0.27107061503416857,
      "grad_norm": 0.0,
      "learning_rate": 1.9691269944221044e-05,
      "loss": 1.8039,
      "step": 119
    },
    {
      "epoch": 0.2733485193621868,
      "grad_norm": 0.0,
      "learning_rate": 1.968831168831169e-05,
      "loss": 2.0044,
      "step": 120
    },
    {
      "epoch": 0.275626423690205,
      "grad_norm": 0.0,
      "learning_rate": 1.9685346508906514e-05,
      "loss": 1.8278,
      "step": 121
    },
    {
      "epoch": 0.27790432801822323,
      "grad_norm": 0.0,
      "learning_rate": 1.9682374381671442e-05,
      "loss": 1.7364,
      "step": 122
    },
    {
      "epoch": 0.28018223234624146,
      "grad_norm": 0.0,
      "learning_rate": 1.9679395282158218e-05,
      "loss": 2.0259,
      "step": 123
    },
    {
      "epoch": 0.2824601366742597,
      "grad_norm": 0.0,
      "learning_rate": 1.967640918580376e-05,
      "loss": 1.7846,
      "step": 124
    },
    {
      "epoch": 0.2847380410022779,
      "grad_norm": 0.0,
      "learning_rate": 1.967341606792946e-05,
      "loss": 1.8559,
      "step": 125
    },
    {
      "epoch": 0.2870159453302961,
      "grad_norm": 0.0,
      "learning_rate": 1.967041590374052e-05,
      "loss": 1.866,
      "step": 126
    },
    {
      "epoch": 0.28929384965831434,
      "grad_norm": 0.0,
      "learning_rate": 1.966740866832526e-05,
      "loss": 1.6675,
      "step": 127
    },
    {
      "epoch": 0.29157175398633256,
      "grad_norm": 0.0,
      "learning_rate": 1.9664394336654432e-05,
      "loss": 1.7645,
      "step": 128
    },
    {
      "epoch": 0.2938496583143508,
      "grad_norm": 0.0,
      "learning_rate": 1.9661372883580523e-05,
      "loss": 1.9748,
      "step": 129
    },
    {
      "epoch": 0.296127562642369,
      "grad_norm": 0.0,
      "learning_rate": 1.9658344283837056e-05,
      "loss": 1.8305,
      "step": 130
    },
    {
      "epoch": 0.2984054669703872,
      "grad_norm": 0.0,
      "learning_rate": 1.965530851203789e-05,
      "loss": 1.9718,
      "step": 131
    },
    {
      "epoch": 0.30068337129840544,
      "grad_norm": 0.0,
      "learning_rate": 1.9652265542676502e-05,
      "loss": 1.7814,
      "step": 132
    },
    {
      "epoch": 0.30296127562642367,
      "grad_norm": 0.0,
      "learning_rate": 1.9649215350125283e-05,
      "loss": 1.9832,
      "step": 133
    },
    {
      "epoch": 0.3052391799544419,
      "grad_norm": 0.0,
      "learning_rate": 1.9646157908634805e-05,
      "loss": 1.6894,
      "step": 134
    },
    {
      "epoch": 0.30751708428246016,
      "grad_norm": 0.0,
      "learning_rate": 1.9643093192333114e-05,
      "loss": 1.86,
      "step": 135
    },
    {
      "epoch": 0.3097949886104784,
      "grad_norm": 0.0,
      "learning_rate": 1.9640021175224987e-05,
      "loss": 1.72,
      "step": 136
    },
    {
      "epoch": 0.3120728929384966,
      "grad_norm": 0.0,
      "learning_rate": 1.9636941831191202e-05,
      "loss": 1.8442,
      "step": 137
    },
    {
      "epoch": 0.3143507972665148,
      "grad_norm": 0.0,
      "learning_rate": 1.9633855133987797e-05,
      "loss": 1.9141,
      "step": 138
    },
    {
      "epoch": 0.31662870159453305,
      "grad_norm": 0.0,
      "learning_rate": 1.963076105724532e-05,
      "loss": 1.8632,
      "step": 139
    },
    {
      "epoch": 0.31890660592255127,
      "grad_norm": 0.0,
      "learning_rate": 1.962765957446809e-05,
      "loss": 1.823,
      "step": 140
    },
    {
      "epoch": 0.3211845102505695,
      "grad_norm": 0.0,
      "learning_rate": 1.962455065903342e-05,
      "loss": 1.9247,
      "step": 141
    },
    {
      "epoch": 0.3234624145785877,
      "grad_norm": 0.0,
      "learning_rate": 1.9621434284190885e-05,
      "loss": 1.9465,
      "step": 142
    },
    {
      "epoch": 0.32574031890660593,
      "grad_norm": 0.0,
      "learning_rate": 1.9618310423061524e-05,
      "loss": 1.8873,
      "step": 143
    },
    {
      "epoch": 0.32801822323462415,
      "grad_norm": 0.0,
      "learning_rate": 1.9615179048637097e-05,
      "loss": 1.9864,
      "step": 144
    },
    {
      "epoch": 0.33029612756264237,
      "grad_norm": 0.0,
      "learning_rate": 1.9612040133779266e-05,
      "loss": 1.9989,
      "step": 145
    },
    {
      "epoch": 0.3325740318906606,
      "grad_norm": 0.0,
      "learning_rate": 1.960889365121886e-05,
      "loss": 1.7546,
      "step": 146
    },
    {
      "epoch": 0.3348519362186788,
      "grad_norm": 0.0,
      "learning_rate": 1.9605739573555053e-05,
      "loss": 1.8092,
      "step": 147
    },
    {
      "epoch": 0.33712984054669703,
      "grad_norm": 0.0,
      "learning_rate": 1.9602577873254568e-05,
      "loss": 1.8627,
      "step": 148
    },
    {
      "epoch": 0.33940774487471526,
      "grad_norm": 0.0,
      "learning_rate": 1.9599408522650895e-05,
      "loss": 1.8892,
      "step": 149
    },
    {
      "epoch": 0.3416856492027335,
      "grad_norm": 0.0,
      "learning_rate": 1.9596231493943474e-05,
      "loss": 1.8105,
      "step": 150
    },
    {
      "epoch": 0.3439635535307517,
      "grad_norm": 0.0,
      "learning_rate": 1.9593046759196877e-05,
      "loss": 1.8469,
      "step": 151
    },
    {
      "epoch": 0.3462414578587699,
      "grad_norm": 0.0,
      "learning_rate": 1.958985429033999e-05,
      "loss": 1.8693,
      "step": 152
    },
    {
      "epoch": 0.34851936218678814,
      "grad_norm": 0.0,
      "learning_rate": 1.9586654059165205e-05,
      "loss": 1.8949,
      "step": 153
    },
    {
      "epoch": 0.35079726651480636,
      "grad_norm": 0.0,
      "learning_rate": 1.9583446037327565e-05,
      "loss": 1.8999,
      "step": 154
    },
    {
      "epoch": 0.3530751708428246,
      "grad_norm": 0.0,
      "learning_rate": 1.9580230196343943e-05,
      "loss": 1.8187,
      "step": 155
    },
    {
      "epoch": 0.3553530751708428,
      "grad_norm": 0.0,
      "learning_rate": 1.957700650759219e-05,
      "loss": 1.68,
      "step": 156
    },
    {
      "epoch": 0.357630979498861,
      "grad_norm": 0.0,
      "learning_rate": 1.9573774942310304e-05,
      "loss": 1.7054,
      "step": 157
    },
    {
      "epoch": 0.35990888382687924,
      "grad_norm": 0.0,
      "learning_rate": 1.9570535471595545e-05,
      "loss": 1.8753,
      "step": 158
    },
    {
      "epoch": 0.3621867881548975,
      "grad_norm": 0.0,
      "learning_rate": 1.9567288066403597e-05,
      "loss": 1.8248,
      "step": 159
    },
    {
      "epoch": 0.36446469248291574,
      "grad_norm": 0.0,
      "learning_rate": 1.9564032697547687e-05,
      "loss": 1.8118,
      "step": 160
    },
    {
      "epoch": 0.36674259681093396,
      "grad_norm": 0.0,
      "learning_rate": 1.9560769335697723e-05,
      "loss": 1.8175,
      "step": 161
    },
    {
      "epoch": 0.3690205011389522,
      "grad_norm": 0.0,
      "learning_rate": 1.9557497951379406e-05,
      "loss": 1.8191,
      "step": 162
    },
    {
      "epoch": 0.3712984054669704,
      "grad_norm": 0.0,
      "learning_rate": 1.9554218514973334e-05,
      "loss": 1.8828,
      "step": 163
    },
    {
      "epoch": 0.3735763097949886,
      "grad_norm": 0.0,
      "learning_rate": 1.955093099671413e-05,
      "loss": 1.9217,
      "step": 164
    },
    {
      "epoch": 0.37585421412300685,
      "grad_norm": 0.0,
      "learning_rate": 1.9547635366689516e-05,
      "loss": 1.9383,
      "step": 165
    },
    {
      "epoch": 0.37813211845102507,
      "grad_norm": 0.0,
      "learning_rate": 1.954433159483942e-05,
      "loss": 1.9053,
      "step": 166
    },
    {
      "epoch": 0.3804100227790433,
      "grad_norm": 0.0,
      "learning_rate": 1.9541019650955066e-05,
      "loss": 1.7693,
      "step": 167
    },
    {
      "epoch": 0.3826879271070615,
      "grad_norm": 0.0,
      "learning_rate": 1.9537699504678044e-05,
      "loss": 1.8591,
      "step": 168
    },
    {
      "epoch": 0.38496583143507973,
      "grad_norm": 0.0,
      "learning_rate": 1.953437112549938e-05,
      "loss": 1.9008,
      "step": 169
    },
    {
      "epoch": 0.38724373576309795,
      "grad_norm": 0.0,
      "learning_rate": 1.953103448275862e-05,
      "loss": 1.7367,
      "step": 170
    },
    {
      "epoch": 0.3895216400911162,
      "grad_norm": 0.0,
      "learning_rate": 1.952768954564287e-05,
      "loss": 1.9731,
      "step": 171
    },
    {
      "epoch": 0.3917995444191344,
      "grad_norm": 0.0,
      "learning_rate": 1.9524336283185842e-05,
      "loss": 1.8188,
      "step": 172
    },
    {
      "epoch": 0.3940774487471526,
      "grad_norm": 0.0,
      "learning_rate": 1.9520974664266928e-05,
      "loss": 1.8901,
      "step": 173
    },
    {
      "epoch": 0.39635535307517084,
      "grad_norm": 0.0,
      "learning_rate": 1.9517604657610205e-05,
      "loss": 1.733,
      "step": 174
    },
    {
      "epoch": 0.39863325740318906,
      "grad_norm": 0.0,
      "learning_rate": 1.9514226231783483e-05,
      "loss": 1.9324,
      "step": 175
    },
    {
      "epoch": 0.4009111617312073,
      "grad_norm": 0.0,
      "learning_rate": 1.9510839355197333e-05,
      "loss": 2.0288,
      "step": 176
    },
    {
      "epoch": 0.4031890660592255,
      "grad_norm": 0.0,
      "learning_rate": 1.9507443996104078e-05,
      "loss": 1.8382,
      "step": 177
    },
    {
      "epoch": 0.4054669703872437,
      "grad_norm": 0.0,
      "learning_rate": 1.9504040122596826e-05,
      "loss": 1.7701,
      "step": 178
    },
    {
      "epoch": 0.40774487471526194,
      "grad_norm": 0.0,
      "learning_rate": 1.9500627702608454e-05,
      "loss": 1.8828,
      "step": 179
    },
    {
      "epoch": 0.41002277904328016,
      "grad_norm": 0.0,
      "learning_rate": 1.9497206703910617e-05,
      "loss": 1.7847,
      "step": 180
    },
    {
      "epoch": 0.4123006833712984,
      "grad_norm": 0.0,
      "learning_rate": 1.9493777094112713e-05,
      "loss": 1.9039,
      "step": 181
    },
    {
      "epoch": 0.4145785876993166,
      "grad_norm": 0.0,
      "learning_rate": 1.949033884066088e-05,
      "loss": 1.8635,
      "step": 182
    },
    {
      "epoch": 0.4168564920273349,
      "grad_norm": 0.0,
      "learning_rate": 1.9486891910836954e-05,
      "loss": 1.671,
      "step": 183
    },
    {
      "epoch": 0.4191343963553531,
      "grad_norm": 0.0,
      "learning_rate": 1.9483436271757442e-05,
      "loss": 2.0151,
      "step": 184
    },
    {
      "epoch": 0.4214123006833713,
      "grad_norm": 0.0,
      "learning_rate": 1.9479971890372456e-05,
      "loss": 1.8151,
      "step": 185
    },
    {
      "epoch": 0.42369020501138954,
      "grad_norm": 0.0,
      "learning_rate": 1.947649873346468e-05,
      "loss": 1.9972,
      "step": 186
    },
    {
      "epoch": 0.42596810933940776,
      "grad_norm": 0.0,
      "learning_rate": 1.9473016767648306e-05,
      "loss": 1.8369,
      "step": 187
    },
    {
      "epoch": 0.428246013667426,
      "grad_norm": 0.0,
      "learning_rate": 1.9469525959367946e-05,
      "loss": 1.7784,
      "step": 188
    },
    {
      "epoch": 0.4305239179954442,
      "grad_norm": 0.0,
      "learning_rate": 1.9466026274897585e-05,
      "loss": 1.7226,
      "step": 189
    },
    {
      "epoch": 0.4328018223234624,
      "grad_norm": 0.0,
      "learning_rate": 1.9462517680339464e-05,
      "loss": 1.8151,
      "step": 190
    },
    {
      "epoch": 0.43507972665148065,
      "grad_norm": 0.0,
      "learning_rate": 1.9459000141623002e-05,
      "loss": 1.8637,
      "step": 191
    },
    {
      "epoch": 0.43735763097949887,
      "grad_norm": 0.0,
      "learning_rate": 1.945547362450369e-05,
      "loss": 1.8807,
      "step": 192
    },
    {
      "epoch": 0.4396355353075171,
      "grad_norm": 0.0,
      "learning_rate": 1.945193809456198e-05,
      "loss": 1.8704,
      "step": 193
    },
    {
      "epoch": 0.4419134396355353,
      "grad_norm": 0.0,
      "learning_rate": 1.9448393517202162e-05,
      "loss": 1.8055,
      "step": 194
    },
    {
      "epoch": 0.44419134396355353,
      "grad_norm": 0.0,
      "learning_rate": 1.9444839857651248e-05,
      "loss": 1.7756,
      "step": 195
    },
    {
      "epoch": 0.44646924829157175,
      "grad_norm": 0.0,
      "learning_rate": 1.9441277080957813e-05,
      "loss": 1.8105,
      "step": 196
    },
    {
      "epoch": 0.44874715261959,
      "grad_norm": 0.0,
      "learning_rate": 1.9437705151990866e-05,
      "loss": 1.9053,
      "step": 197
    },
    {
      "epoch": 0.4510250569476082,
      "grad_norm": 0.0,
      "learning_rate": 1.94341240354387e-05,
      "loss": 1.975,
      "step": 198
    },
    {
      "epoch": 0.4533029612756264,
      "grad_norm": 0.0,
      "learning_rate": 1.94305336958077e-05,
      "loss": 1.7839,
      "step": 199
    },
    {
      "epoch": 0.45558086560364464,
      "grad_norm": 0.0,
      "learning_rate": 1.9426934097421205e-05,
      "loss": 1.765,
      "step": 200
    },
    {
      "epoch": 0.45785876993166286,
      "grad_norm": 0.0,
      "learning_rate": 1.9423325204418306e-05,
      "loss": 1.9151,
      "step": 201
    },
    {
      "epoch": 0.4601366742596811,
      "grad_norm": 0.0,
      "learning_rate": 1.941970698075266e-05,
      "loss": 2.0266,
      "step": 202
    },
    {
      "epoch": 0.4624145785876993,
      "grad_norm": 0.0,
      "learning_rate": 1.9416079390191285e-05,
      "loss": 1.8082,
      "step": 203
    },
    {
      "epoch": 0.4646924829157175,
      "grad_norm": 0.0,
      "learning_rate": 1.9412442396313366e-05,
      "loss": 1.8551,
      "step": 204
    },
    {
      "epoch": 0.46697038724373574,
      "grad_norm": 0.0,
      "learning_rate": 1.9408795962509016e-05,
      "loss": 1.7932,
      "step": 205
    },
    {
      "epoch": 0.46924829157175396,
      "grad_norm": 0.0,
      "learning_rate": 1.9405140051978056e-05,
      "loss": 1.7762,
      "step": 206
    },
    {
      "epoch": 0.4715261958997722,
      "grad_norm": 0.0,
      "learning_rate": 1.9401474627728788e-05,
      "loss": 1.9933,
      "step": 207
    },
    {
      "epoch": 0.47380410022779046,
      "grad_norm": 0.0,
      "learning_rate": 1.9397799652576727e-05,
      "loss": 1.9953,
      "step": 208
    },
    {
      "epoch": 0.4760820045558087,
      "grad_norm": 0.0,
      "learning_rate": 1.9394115089143356e-05,
      "loss": 1.7761,
      "step": 209
    },
    {
      "epoch": 0.4783599088838269,
      "grad_norm": 0.0,
      "learning_rate": 1.939042089985486e-05,
      "loss": 1.6961,
      "step": 210
    },
    {
      "epoch": 0.4806378132118451,
      "grad_norm": 0.0,
      "learning_rate": 1.9386717046940853e-05,
      "loss": 1.7874,
      "step": 211
    },
    {
      "epoch": 0.48291571753986334,
      "grad_norm": 0.0,
      "learning_rate": 1.9383003492433062e-05,
      "loss": 2.0687,
      "step": 212
    },
    {
      "epoch": 0.48519362186788156,
      "grad_norm": 0.0,
      "learning_rate": 1.937928019816407e-05,
      "loss": 1.8469,
      "step": 213
    },
    {
      "epoch": 0.4874715261958998,
      "grad_norm": 0.0,
      "learning_rate": 1.9375547125765978e-05,
      "loss": 1.861,
      "step": 214
    },
    {
      "epoch": 0.489749430523918,
      "grad_norm": 0.0,
      "learning_rate": 1.9371804236669104e-05,
      "loss": 1.8835,
      "step": 215
    },
    {
      "epoch": 0.4920273348519362,
      "grad_norm": 0.0,
      "learning_rate": 1.9368051492100644e-05,
      "loss": 1.8315,
      "step": 216
    },
    {
      "epoch": 0.49430523917995445,
      "grad_norm": 0.0,
      "learning_rate": 1.9364288853083345e-05,
      "loss": 1.9031,
      "step": 217
    },
    {
      "epoch": 0.49658314350797267,
      "grad_norm": 0.0,
      "learning_rate": 1.9360516280434146e-05,
      "loss": 1.9127,
      "step": 218
    },
    {
      "epoch": 0.4988610478359909,
      "grad_norm": 0.0,
      "learning_rate": 1.9356733734762817e-05,
      "loss": 1.7706,
      "step": 219
    },
    {
      "epoch": 0.5011389521640092,
      "grad_norm": 0.0,
      "learning_rate": 1.935294117647059e-05,
      "loss": 1.9683,
      "step": 220
    },
    {
      "epoch": 0.5034168564920274,
      "grad_norm": 0.0,
      "learning_rate": 1.9349138565748787e-05,
      "loss": 1.6735,
      "step": 221
    },
    {
      "epoch": 0.5056947608200456,
      "grad_norm": 0.0,
      "learning_rate": 1.9345325862577412e-05,
      "loss": 1.8115,
      "step": 222
    },
    {
      "epoch": 0.5079726651480638,
      "grad_norm": 0.0,
      "learning_rate": 1.9341503026723757e-05,
      "loss": 1.8155,
      "step": 223
    },
    {
      "epoch": 0.510250569476082,
      "grad_norm": 0.0,
      "learning_rate": 1.933767001774098e-05,
      "loss": 1.8103,
      "step": 224
    },
    {
      "epoch": 0.5125284738041003,
      "grad_norm": 0.0,
      "learning_rate": 1.9333826794966695e-05,
      "loss": 1.7367,
      "step": 225
    },
    {
      "epoch": 0.5148063781321185,
      "grad_norm": 0.0,
      "learning_rate": 1.9329973317521497e-05,
      "loss": 1.7631,
      "step": 226
    },
    {
      "epoch": 0.5170842824601367,
      "grad_norm": 0.0,
      "learning_rate": 1.9326109544307555e-05,
      "loss": 1.8696,
      "step": 227
    },
    {
      "epoch": 0.5193621867881549,
      "grad_norm": 0.0,
      "learning_rate": 1.9322235434007134e-05,
      "loss": 1.8613,
      "step": 228
    },
    {
      "epoch": 0.5216400911161732,
      "grad_norm": 0.0,
      "learning_rate": 1.9318350945081115e-05,
      "loss": 1.6882,
      "step": 229
    },
    {
      "epoch": 0.5239179954441914,
      "grad_norm": 0.0,
      "learning_rate": 1.9314456035767514e-05,
      "loss": 1.8036,
      "step": 230
    },
    {
      "epoch": 0.5261958997722096,
      "grad_norm": 0.0,
      "learning_rate": 1.9310550664079988e-05,
      "loss": 1.7296,
      "step": 231
    },
    {
      "epoch": 0.5284738041002278,
      "grad_norm": 0.0,
      "learning_rate": 1.930663478780634e-05,
      "loss": 1.6914,
      "step": 232
    },
    {
      "epoch": 0.530751708428246,
      "grad_norm": 0.0,
      "learning_rate": 1.930270836450696e-05,
      "loss": 1.9104,
      "step": 233
    },
    {
      "epoch": 0.5330296127562643,
      "grad_norm": 0.0,
      "learning_rate": 1.9298771351513336e-05,
      "loss": 1.7403,
      "step": 234
    },
    {
      "epoch": 0.5353075170842825,
      "grad_norm": 0.0,
      "learning_rate": 1.9294823705926484e-05,
      "loss": 1.8471,
      "step": 235
    },
    {
      "epoch": 0.5375854214123007,
      "grad_norm": 0.0,
      "learning_rate": 1.9290865384615384e-05,
      "loss": 1.8498,
      "step": 236
    },
    {
      "epoch": 0.5398633257403189,
      "grad_norm": 0.0,
      "learning_rate": 1.9286896344215435e-05,
      "loss": 1.8415,
      "step": 237
    },
    {
      "epoch": 0.5421412300683371,
      "grad_norm": 0.0,
      "learning_rate": 1.9282916541126848e-05,
      "loss": 1.868,
      "step": 238
    },
    {
      "epoch": 0.5444191343963554,
      "grad_norm": 0.0,
      "learning_rate": 1.927892593151305e-05,
      "loss": 1.8843,
      "step": 239
    },
    {
      "epoch": 0.5466970387243736,
      "grad_norm": 0.0,
      "learning_rate": 1.9274924471299096e-05,
      "loss": 1.9664,
      "step": 240
    },
    {
      "epoch": 0.5489749430523918,
      "grad_norm": 0.0,
      "learning_rate": 1.927091211617002e-05,
      "loss": 1.6881,
      "step": 241
    },
    {
      "epoch": 0.55125284738041,
      "grad_norm": 0.0,
      "learning_rate": 1.9266888821569223e-05,
      "loss": 1.7565,
      "step": 242
    },
    {
      "epoch": 0.5535307517084282,
      "grad_norm": 0.0,
      "learning_rate": 1.92628545426968e-05,
      "loss": 1.7555,
      "step": 243
    },
    {
      "epoch": 0.5558086560364465,
      "grad_norm": 0.0,
      "learning_rate": 1.92588092345079e-05,
      "loss": 1.719,
      "step": 244
    },
    {
      "epoch": 0.5580865603644647,
      "grad_norm": 0.0,
      "learning_rate": 1.9254752851711028e-05,
      "loss": 1.8767,
      "step": 245
    },
    {
      "epoch": 0.5603644646924829,
      "grad_norm": 0.0,
      "learning_rate": 1.9250685348766376e-05,
      "loss": 1.8385,
      "step": 246
    },
    {
      "epoch": 0.5626423690205011,
      "grad_norm": 0.0,
      "learning_rate": 1.9246606679884093e-05,
      "loss": 1.8968,
      "step": 247
    },
    {
      "epoch": 0.5649202733485194,
      "grad_norm": 0.0,
      "learning_rate": 1.9242516799022603e-05,
      "loss": 1.6632,
      "step": 248
    },
    {
      "epoch": 0.5671981776765376,
      "grad_norm": 0.0,
      "learning_rate": 1.9238415659886835e-05,
      "loss": 1.9958,
      "step": 249
    },
    {
      "epoch": 0.5694760820045558,
      "grad_norm": 0.0,
      "learning_rate": 1.9234303215926494e-05,
      "loss": 1.7144,
      "step": 250
    },
    {
      "epoch": 0.571753986332574,
      "grad_norm": 0.0,
      "learning_rate": 1.9230179420334307e-05,
      "loss": 1.8842,
      "step": 251
    },
    {
      "epoch": 0.5740318906605922,
      "grad_norm": 0.0,
      "learning_rate": 1.922604422604423e-05,
      "loss": 1.7184,
      "step": 252
    },
    {
      "epoch": 0.5763097949886105,
      "grad_norm": 0.0,
      "learning_rate": 1.9221897585729664e-05,
      "loss": 1.8718,
      "step": 253
    },
    {
      "epoch": 0.5785876993166287,
      "grad_norm": 0.0,
      "learning_rate": 1.9217739451801666e-05,
      "loss": 1.726,
      "step": 254
    },
    {
      "epoch": 0.5808656036446469,
      "grad_norm": 0.0,
      "learning_rate": 1.9213569776407097e-05,
      "loss": 1.7056,
      "step": 255
    },
    {
      "epoch": 0.5831435079726651,
      "grad_norm": 0.0,
      "learning_rate": 1.9209388511426808e-05,
      "loss": 1.7493,
      "step": 256
    },
    {
      "epoch": 0.5854214123006833,
      "grad_norm": 0.0,
      "learning_rate": 1.920519560847379e-05,
      "loss": 1.8129,
      "step": 257
    },
    {
      "epoch": 0.5876993166287016,
      "grad_norm": 0.0,
      "learning_rate": 1.9200991018891298e-05,
      "loss": 1.8607,
      "step": 258
    },
    {
      "epoch": 0.5899772209567198,
      "grad_norm": 0.0,
      "learning_rate": 1.9196774693750973e-05,
      "loss": 1.7757,
      "step": 259
    },
    {
      "epoch": 0.592255125284738,
      "grad_norm": 0.0,
      "learning_rate": 1.9192546583850936e-05,
      "loss": 1.7289,
      "step": 260
    },
    {
      "epoch": 0.5945330296127562,
      "grad_norm": 0.0,
      "learning_rate": 1.918830663971389e-05,
      "loss": 1.7827,
      "step": 261
    },
    {
      "epoch": 0.5968109339407744,
      "grad_norm": 0.0,
      "learning_rate": 1.9184054811585176e-05,
      "loss": 1.8738,
      "step": 262
    },
    {
      "epoch": 0.5990888382687927,
      "grad_norm": 0.0,
      "learning_rate": 1.9179791049430845e-05,
      "loss": 1.7512,
      "step": 263
    },
    {
      "epoch": 0.6013667425968109,
      "grad_norm": 0.0,
      "learning_rate": 1.9175515302935666e-05,
      "loss": 1.9041,
      "step": 264
    },
    {
      "epoch": 0.6036446469248291,
      "grad_norm": 0.0,
      "learning_rate": 1.9171227521501178e-05,
      "loss": 1.761,
      "step": 265
    },
    {
      "epoch": 0.6059225512528473,
      "grad_norm": 0.0,
      "learning_rate": 1.9166927654243663e-05,
      "loss": 1.8059,
      "step": 266
    },
    {
      "epoch": 0.6082004555808656,
      "grad_norm": 0.0,
      "learning_rate": 1.916261564999216e-05,
      "loss": 1.8029,
      "step": 267
    },
    {
      "epoch": 0.6104783599088838,
      "grad_norm": 0.0,
      "learning_rate": 1.915829145728643e-05,
      "loss": 1.7574,
      "step": 268
    },
    {
      "epoch": 0.6127562642369021,
      "grad_norm": 0.0,
      "learning_rate": 1.9153955024374905e-05,
      "loss": 1.7662,
      "step": 269
    },
    {
      "epoch": 0.6150341685649203,
      "grad_norm": 0.0,
      "learning_rate": 1.91496062992126e-05,
      "loss": 1.8448,
      "step": 270
    },
    {
      "epoch": 0.6173120728929385,
      "grad_norm": 0.0,
      "learning_rate": 1.9145245229459078e-05,
      "loss": 1.6852,
      "step": 271
    },
    {
      "epoch": 0.6195899772209568,
      "grad_norm": 0.0,
      "learning_rate": 1.9140871762476314e-05,
      "loss": 1.773,
      "step": 272
    },
    {
      "epoch": 0.621867881548975,
      "grad_norm": 0.0,
      "learning_rate": 1.913648584532659e-05,
      "loss": 1.8345,
      "step": 273
    },
    {
      "epoch": 0.6241457858769932,
      "grad_norm": 0.0,
      "learning_rate": 1.9132087424770354e-05,
      "loss": 1.7964,
      "step": 274
    },
    {
      "epoch": 0.6264236902050114,
      "grad_norm": 0.0,
      "learning_rate": 1.912767644726408e-05,
      "loss": 1.8436,
      "step": 275
    },
    {
      "epoch": 0.6287015945330297,
      "grad_norm": 0.0,
      "learning_rate": 1.912325285895807e-05,
      "loss": 1.9349,
      "step": 276
    },
    {
      "epoch": 0.6309794988610479,
      "grad_norm": 0.0,
      "learning_rate": 1.911881660569429e-05,
      "loss": 1.7675,
      "step": 277
    },
    {
      "epoch": 0.6332574031890661,
      "grad_norm": 0.0,
      "learning_rate": 1.911436763300414e-05,
      "loss": 1.8384,
      "step": 278
    },
    {
      "epoch": 0.6355353075170843,
      "grad_norm": 0.0,
      "learning_rate": 1.910990588610624e-05,
      "loss": 1.8475,
      "step": 279
    },
    {
      "epoch": 0.6378132118451025,
      "grad_norm": 0.0,
      "learning_rate": 1.9105431309904158e-05,
      "loss": 1.852,
      "step": 280
    },
    {
      "epoch": 0.6400911161731208,
      "grad_norm": 0.0,
      "learning_rate": 1.9100943848984165e-05,
      "loss": 1.7185,
      "step": 281
    },
    {
      "epoch": 0.642369020501139,
      "grad_norm": 0.0,
      "learning_rate": 1.9096443447612948e-05,
      "loss": 1.7923,
      "step": 282
    },
    {
      "epoch": 0.6446469248291572,
      "grad_norm": 0.0,
      "learning_rate": 1.9091930049735282e-05,
      "loss": 1.9492,
      "step": 283
    },
    {
      "epoch": 0.6469248291571754,
      "grad_norm": 0.0,
      "learning_rate": 1.9087403598971723e-05,
      "loss": 1.9283,
      "step": 284
    },
    {
      "epoch": 0.6492027334851936,
      "grad_norm": 0.0,
      "learning_rate": 1.9082864038616254e-05,
      "loss": 1.8545,
      "step": 285
    },
    {
      "epoch": 0.6514806378132119,
      "grad_norm": 0.0,
      "learning_rate": 1.9078311311633904e-05,
      "loss": 1.8646,
      "step": 286
    },
    {
      "epoch": 0.6537585421412301,
      "grad_norm": 0.0,
      "learning_rate": 1.9073745360658385e-05,
      "loss": 1.8924,
      "step": 287
    },
    {
      "epoch": 0.6560364464692483,
      "grad_norm": 0.0,
      "learning_rate": 1.906916612798966e-05,
      "loss": 1.8022,
      "step": 288
    },
    {
      "epoch": 0.6583143507972665,
      "grad_norm": 0.0,
      "learning_rate": 1.9064573555591522e-05,
      "loss": 1.8965,
      "step": 289
    },
    {
      "epoch": 0.6605922551252847,
      "grad_norm": 0.0,
      "learning_rate": 1.9059967585089144e-05,
      "loss": 1.9303,
      "step": 290
    },
    {
      "epoch": 0.662870159453303,
      "grad_norm": 0.0,
      "learning_rate": 1.9055348157766597e-05,
      "loss": 1.8566,
      "step": 291
    },
    {
      "epoch": 0.6651480637813212,
      "grad_norm": 0.0,
      "learning_rate": 1.9050715214564374e-05,
      "loss": 1.7818,
      "step": 292
    },
    {
      "epoch": 0.6674259681093394,
      "grad_norm": 0.0,
      "learning_rate": 1.904606869607684e-05,
      "loss": 1.9262,
      "step": 293
    },
    {
      "epoch": 0.6697038724373576,
      "grad_norm": 0.0,
      "learning_rate": 1.9041408542549725e-05,
      "loss": 1.6944,
      "step": 294
    },
    {
      "epoch": 0.6719817767653758,
      "grad_norm": 0.0,
      "learning_rate": 1.9036734693877552e-05,
      "loss": 1.7894,
      "step": 295
    },
    {
      "epoch": 0.6742596810933941,
      "grad_norm": 0.0,
      "learning_rate": 1.9032047089601048e-05,
      "loss": 1.9153,
      "step": 296
    },
    {
      "epoch": 0.6765375854214123,
      "grad_norm": 0.0,
      "learning_rate": 1.9027345668904537e-05,
      "loss": 1.8007,
      "step": 297
    },
    {
      "epoch": 0.6788154897494305,
      "grad_norm": 0.0,
      "learning_rate": 1.9022630370613315e-05,
      "loss": 1.8416,
      "step": 298
    },
    {
      "epoch": 0.6810933940774487,
      "grad_norm": 0.0,
      "learning_rate": 1.9017901133191004e-05,
      "loss": 1.6625,
      "step": 299
    },
    {
      "epoch": 0.683371298405467,
      "grad_norm": 0.0,
      "learning_rate": 1.9013157894736845e-05,
      "loss": 1.8183,
      "step": 300
    },
    {
      "epoch": 0.6856492027334852,
      "grad_norm": 0.0,
      "learning_rate": 1.9008400592983036e-05,
      "loss": 1.8195,
      "step": 301
    },
    {
      "epoch": 0.6879271070615034,
      "grad_norm": 0.0,
      "learning_rate": 1.9003629165291983e-05,
      "loss": 1.7959,
      "step": 302
    },
    {
      "epoch": 0.6902050113895216,
      "grad_norm": 0.0,
      "learning_rate": 1.899884354865356e-05,
      "loss": 1.7854,
      "step": 303
    },
    {
      "epoch": 0.6924829157175398,
      "grad_norm": 0.0,
      "learning_rate": 1.899404367968233e-05,
      "loss": 1.7821,
      "step": 304
    },
    {
      "epoch": 0.6947608200455581,
      "grad_norm": 0.0,
      "learning_rate": 1.8989229494614746e-05,
      "loss": 1.9069,
      "step": 305
    },
    {
      "epoch": 0.6970387243735763,
      "grad_norm": 0.0,
      "learning_rate": 1.8984400929306342e-05,
      "loss": 1.7895,
      "step": 306
    },
    {
      "epoch": 0.6993166287015945,
      "grad_norm": 0.0,
      "learning_rate": 1.8979557919228855e-05,
      "loss": 1.818,
      "step": 307
    },
    {
      "epoch": 0.7015945330296127,
      "grad_norm": 0.0,
      "learning_rate": 1.897470039946738e-05,
      "loss": 1.8809,
      "step": 308
    },
    {
      "epoch": 0.7038724373576309,
      "grad_norm": 0.0,
      "learning_rate": 1.8969828304717456e-05,
      "loss": 1.6762,
      "step": 309
    },
    {
      "epoch": 0.7061503416856492,
      "grad_norm": 0.0,
      "learning_rate": 1.8964941569282137e-05,
      "loss": 1.7813,
      "step": 310
    },
    {
      "epoch": 0.7084282460136674,
      "grad_norm": 0.0,
      "learning_rate": 1.896004012706905e-05,
      "loss": 1.8738,
      "step": 311
    },
    {
      "epoch": 0.7107061503416856,
      "grad_norm": 0.0,
      "learning_rate": 1.8955123911587412e-05,
      "loss": 1.7677,
      "step": 312
    },
    {
      "epoch": 0.7129840546697038,
      "grad_norm": 0.0,
      "learning_rate": 1.8950192855945e-05,
      "loss": 1.7614,
      "step": 313
    },
    {
      "epoch": 0.715261958997722,
      "grad_norm": 0.0,
      "learning_rate": 1.894524689284515e-05,
      "loss": 1.9156,
      "step": 314
    },
    {
      "epoch": 0.7175398633257403,
      "grad_norm": 0.0,
      "learning_rate": 1.8940285954583686e-05,
      "loss": 1.6247,
      "step": 315
    },
    {
      "epoch": 0.7198177676537585,
      "grad_norm": 0.0,
      "learning_rate": 1.8935309973045825e-05,
      "loss": 1.7746,
      "step": 316
    },
    {
      "epoch": 0.7220956719817767,
      "grad_norm": 0.0,
      "learning_rate": 1.8930318879703055e-05,
      "loss": 1.6941,
      "step": 317
    },
    {
      "epoch": 0.724373576309795,
      "grad_norm": 0.0,
      "learning_rate": 1.8925312605610003e-05,
      "loss": 1.7349,
      "step": 318
    },
    {
      "epoch": 0.7266514806378133,
      "grad_norm": 0.0,
      "learning_rate": 1.8920291081401257e-05,
      "loss": 1.786,
      "step": 319
    },
    {
      "epoch": 0.7289293849658315,
      "grad_norm": 0.0,
      "learning_rate": 1.8915254237288138e-05,
      "loss": 1.87,
      "step": 320
    },
    {
      "epoch": 0.7312072892938497,
      "grad_norm": 0.0,
      "learning_rate": 1.891020200305551e-05,
      "loss": 1.8122,
      "step": 321
    },
    {
      "epoch": 0.7334851936218679,
      "grad_norm": 0.0,
      "learning_rate": 1.8905134308058486e-05,
      "loss": 1.9103,
      "step": 322
    },
    {
      "epoch": 0.7357630979498861,
      "grad_norm": 0.0,
      "learning_rate": 1.8900051081219138e-05,
      "loss": 1.7457,
      "step": 323
    },
    {
      "epoch": 0.7380410022779044,
      "grad_norm": 0.0,
      "learning_rate": 1.8894952251023194e-05,
      "loss": 1.7686,
      "step": 324
    },
    {
      "epoch": 0.7403189066059226,
      "grad_norm": 0.0,
      "learning_rate": 1.8889837745516653e-05,
      "loss": 1.7409,
      "step": 325
    },
    {
      "epoch": 0.7425968109339408,
      "grad_norm": 0.0,
      "learning_rate": 1.8884707492302432e-05,
      "loss": 1.9199,
      "step": 326
    },
    {
      "epoch": 0.744874715261959,
      "grad_norm": 0.0,
      "learning_rate": 1.8879561418536924e-05,
      "loss": 1.844,
      "step": 327
    },
    {
      "epoch": 0.7471526195899773,
      "grad_norm": 0.0,
      "learning_rate": 1.8874399450926562e-05,
      "loss": 1.6653,
      "step": 328
    },
    {
      "epoch": 0.7494305239179955,
      "grad_norm": 0.0,
      "learning_rate": 1.886922151572435e-05,
      "loss": 1.9141,
      "step": 329
    },
    {
      "epoch": 0.7517084282460137,
      "grad_norm": 0.0,
      "learning_rate": 1.8864027538726336e-05,
      "loss": 1.8721,
      "step": 330
    },
    {
      "epoch": 0.7539863325740319,
      "grad_norm": 0.0,
      "learning_rate": 1.885881744526806e-05,
      "loss": 1.6069,
      "step": 331
    },
    {
      "epoch": 0.7562642369020501,
      "grad_norm": 0.0,
      "learning_rate": 1.8853591160221e-05,
      "loss": 1.9076,
      "step": 332
    },
    {
      "epoch": 0.7585421412300684,
      "grad_norm": 0.0,
      "learning_rate": 1.8848348607988935e-05,
      "loss": 1.7952,
      "step": 333
    },
    {
      "epoch": 0.7608200455580866,
      "grad_norm": 0.0,
      "learning_rate": 1.884308971250433e-05,
      "loss": 1.6736,
      "step": 334
    },
    {
      "epoch": 0.7630979498861048,
      "grad_norm": 0.0,
      "learning_rate": 1.8837814397224635e-05,
      "loss": 1.9223,
      "step": 335
    },
    {
      "epoch": 0.765375854214123,
      "grad_norm": 0.0,
      "learning_rate": 1.8832522585128564e-05,
      "loss": 1.7733,
      "step": 336
    },
    {
      "epoch": 0.7676537585421412,
      "grad_norm": 0.0,
      "learning_rate": 1.8827214198712374e-05,
      "loss": 1.7597,
      "step": 337
    },
    {
      "epoch": 0.7699316628701595,
      "grad_norm": 0.0,
      "learning_rate": 1.882188915998606e-05,
      "loss": 1.6681,
      "step": 338
    },
    {
      "epoch": 0.7722095671981777,
      "grad_norm": 0.0,
      "learning_rate": 1.8816547390469545e-05,
      "loss": 1.6973,
      "step": 339
    },
    {
      "epoch": 0.7744874715261959,
      "grad_norm": 0.0,
      "learning_rate": 1.8811188811188814e-05,
      "loss": 1.7656,
      "step": 340
    },
    {
      "epoch": 0.7767653758542141,
      "grad_norm": 0.0,
      "learning_rate": 1.880581334267204e-05,
      "loss": 2.0147,
      "step": 341
    },
    {
      "epoch": 0.7790432801822323,
      "grad_norm": 0.0,
      "learning_rate": 1.8800420904945635e-05,
      "loss": 1.7254,
      "step": 342
    },
    {
      "epoch": 0.7813211845102506,
      "grad_norm": 0.0,
      "learning_rate": 1.87950114175303e-05,
      "loss": 1.864,
      "step": 343
    },
    {
      "epoch": 0.7835990888382688,
      "grad_norm": 0.0,
      "learning_rate": 1.8789584799437017e-05,
      "loss": 1.7519,
      "step": 344
    },
    {
      "epoch": 0.785876993166287,
      "grad_norm": 0.0,
      "learning_rate": 1.8784140969162997e-05,
      "loss": 1.8621,
      "step": 345
    },
    {
      "epoch": 0.7881548974943052,
      "grad_norm": 0.0,
      "learning_rate": 1.8778679844687614e-05,
      "loss": 1.8879,
      "step": 346
    },
    {
      "epoch": 0.7904328018223234,
      "grad_norm": 0.0,
      "learning_rate": 1.8773201343468273e-05,
      "loss": 1.6798,
      "step": 347
    },
    {
      "epoch": 0.7927107061503417,
      "grad_norm": 0.0,
      "learning_rate": 1.8767705382436264e-05,
      "loss": 1.9381,
      "step": 348
    },
    {
      "epoch": 0.7949886104783599,
      "grad_norm": 0.0,
      "learning_rate": 1.8762191877992554e-05,
      "loss": 2.0407,
      "step": 349
    },
    {
      "epoch": 0.7972665148063781,
      "grad_norm": 0.0,
      "learning_rate": 1.8756660746003553e-05,
      "loss": 1.7358,
      "step": 350
    },
    {
      "epoch": 0.7995444191343963,
      "grad_norm": 0.0,
      "learning_rate": 1.8751111901796833e-05,
      "loss": 1.8904,
      "step": 351
    },
    {
      "epoch": 0.8018223234624146,
      "grad_norm": 0.0,
      "learning_rate": 1.8745545260156807e-05,
      "loss": 1.8213,
      "step": 352
    },
    {
      "epoch": 0.8041002277904328,
      "grad_norm": 0.0,
      "learning_rate": 1.8739960735320364e-05,
      "loss": 1.9779,
      "step": 353
    },
    {
      "epoch": 0.806378132118451,
      "grad_norm": 0.0,
      "learning_rate": 1.873435824097247e-05,
      "loss": 1.7869,
      "step": 354
    },
    {
      "epoch": 0.8086560364464692,
      "grad_norm": 0.0,
      "learning_rate": 1.872873769024172e-05,
      "loss": 1.7634,
      "step": 355
    },
    {
      "epoch": 0.8109339407744874,
      "grad_norm": 0.0,
      "learning_rate": 1.872309899569584e-05,
      "loss": 1.7391,
      "step": 356
    },
    {
      "epoch": 0.8132118451025057,
      "grad_norm": 0.0,
      "learning_rate": 1.8717442069337165e-05,
      "loss": 1.7673,
      "step": 357
    },
    {
      "epoch": 0.8154897494305239,
      "grad_norm": 0.0,
      "learning_rate": 1.8711766822598056e-05,
      "loss": 1.839,
      "step": 358
    },
    {
      "epoch": 0.8177676537585421,
      "grad_norm": 0.0,
      "learning_rate": 1.870607316633628e-05,
      "loss": 1.693,
      "step": 359
    },
    {
      "epoch": 0.8200455580865603,
      "grad_norm": 0.0,
      "learning_rate": 1.8700361010830327e-05,
      "loss": 1.8105,
      "step": 360
    },
    {
      "epoch": 0.8223234624145785,
      "grad_norm": 0.0,
      "learning_rate": 1.8694630265774728e-05,
      "loss": 1.7951,
      "step": 361
    },
    {
      "epoch": 0.8246013667425968,
      "grad_norm": 0.0,
      "learning_rate": 1.8688880840275266e-05,
      "loss": 1.6912,
      "step": 362
    },
    {
      "epoch": 0.826879271070615,
      "grad_norm": 0.0,
      "learning_rate": 1.8683112642844187e-05,
      "loss": 1.7662,
      "step": 363
    },
    {
      "epoch": 0.8291571753986332,
      "grad_norm": 0.0,
      "learning_rate": 1.8677325581395348e-05,
      "loss": 1.9301,
      "step": 364
    },
    {
      "epoch": 0.8314350797266514,
      "grad_norm": 0.0,
      "learning_rate": 1.867151956323931e-05,
      "loss": 1.7746,
      "step": 365
    },
    {
      "epoch": 0.8337129840546698,
      "grad_norm": 0.0,
      "learning_rate": 1.8665694495078384e-05,
      "loss": 1.8972,
      "step": 366
    },
    {
      "epoch": 0.835990888382688,
      "grad_norm": 0.0,
      "learning_rate": 1.8659850283001647e-05,
      "loss": 1.9447,
      "step": 367
    },
    {
      "epoch": 0.8382687927107062,
      "grad_norm": 0.0,
      "learning_rate": 1.8653986832479884e-05,
      "loss": 1.7172,
      "step": 368
    },
    {
      "epoch": 0.8405466970387244,
      "grad_norm": 0.0,
      "learning_rate": 1.8648104048360505e-05,
      "loss": 1.8894,
      "step": 369
    },
    {
      "epoch": 0.8428246013667426,
      "grad_norm": 0.0,
      "learning_rate": 1.8642201834862387e-05,
      "loss": 1.8513,
      "step": 370
    },
    {
      "epoch": 0.8451025056947609,
      "grad_norm": 0.0,
      "learning_rate": 1.8636280095570668e-05,
      "loss": 1.7918,
      "step": 371
    },
    {
      "epoch": 0.8473804100227791,
      "grad_norm": 0.0,
      "learning_rate": 1.863033873343152e-05,
      "loss": 1.7627,
      "step": 372
    },
    {
      "epoch": 0.8496583143507973,
      "grad_norm": 0.0,
      "learning_rate": 1.8624377650746822e-05,
      "loss": 1.9015,
      "step": 373
    },
    {
      "epoch": 0.8519362186788155,
      "grad_norm": 0.0,
      "learning_rate": 1.8618396749168824e-05,
      "loss": 1.8908,
      "step": 374
    },
    {
      "epoch": 0.8542141230068337,
      "grad_norm": 0.0,
      "learning_rate": 1.8612395929694728e-05,
      "loss": 1.6868,
      "step": 375
    },
    {
      "epoch": 0.856492027334852,
      "grad_norm": 0.0,
      "learning_rate": 1.8606375092661234e-05,
      "loss": 1.8831,
      "step": 376
    },
    {
      "epoch": 0.8587699316628702,
      "grad_norm": 0.0,
      "learning_rate": 1.8600334137739e-05,
      "loss": 1.8061,
      "step": 377
    },
    {
      "epoch": 0.8610478359908884,
      "grad_norm": 0.0,
      "learning_rate": 1.8594272963927112e-05,
      "loss": 1.753,
      "step": 378
    },
    {
      "epoch": 0.8633257403189066,
      "grad_norm": 0.0,
      "learning_rate": 1.8588191469547407e-05,
      "loss": 1.8116,
      "step": 379
    },
    {
      "epoch": 0.8656036446469249,
      "grad_norm": 0.0,
      "learning_rate": 1.858208955223881e-05,
      "loss": 1.8972,
      "step": 380
    },
    {
      "epoch": 0.8678815489749431,
      "grad_norm": 0.0,
      "learning_rate": 1.85759671089516e-05,
      "loss": 1.769,
      "step": 381
    },
    {
      "epoch": 0.8701594533029613,
      "grad_norm": 0.0,
      "learning_rate": 1.85698240359416e-05,
      "loss": 1.677,
      "step": 382
    },
    {
      "epoch": 0.8724373576309795,
      "grad_norm": 0.0,
      "learning_rate": 1.85636602287643e-05,
      "loss": 1.8313,
      "step": 383
    },
    {
      "epoch": 0.8747152619589977,
      "grad_norm": 0.0,
      "learning_rate": 1.8557475582268972e-05,
      "loss": 1.8365,
      "step": 384
    },
    {
      "epoch": 0.876993166287016,
      "grad_norm": 0.0,
      "learning_rate": 1.8551269990592663e-05,
      "loss": 1.7318,
      "step": 385
    },
    {
      "epoch": 0.8792710706150342,
      "grad_norm": 0.0,
      "learning_rate": 1.8545043347154167e-05,
      "loss": 1.7535,
      "step": 386
    },
    {
      "epoch": 0.8815489749430524,
      "grad_norm": 0.0,
      "learning_rate": 1.8538795544647917e-05,
      "loss": 1.8066,
      "step": 387
    },
    {
      "epoch": 0.8838268792710706,
      "grad_norm": 0.0,
      "learning_rate": 1.8532526475037826e-05,
      "loss": 1.8251,
      "step": 388
    },
    {
      "epoch": 0.8861047835990888,
      "grad_norm": 0.0,
      "learning_rate": 1.8526236029551052e-05,
      "loss": 1.7838,
      "step": 389
    },
    {
      "epoch": 0.8883826879271071,
      "grad_norm": 0.0,
      "learning_rate": 1.851992409867173e-05,
      "loss": 1.9,
      "step": 390
    },
    {
      "epoch": 0.8906605922551253,
      "grad_norm": 0.0,
      "learning_rate": 1.8513590572134575e-05,
      "loss": 1.72,
      "step": 391
    },
    {
      "epoch": 0.8929384965831435,
      "grad_norm": 0.0,
      "learning_rate": 1.8507235338918506e-05,
      "loss": 1.7089,
      "step": 392
    },
    {
      "epoch": 0.8952164009111617,
      "grad_norm": 0.0,
      "learning_rate": 1.8500858287240127e-05,
      "loss": 1.864,
      "step": 393
    },
    {
      "epoch": 0.89749430523918,
      "grad_norm": 0.0,
      "learning_rate": 1.8494459304547195e-05,
      "loss": 1.8561,
      "step": 394
    },
    {
      "epoch": 0.8997722095671982,
      "grad_norm": 0.0,
      "learning_rate": 1.8488038277511965e-05,
      "loss": 1.9116,
      "step": 395
    },
    {
      "epoch": 0.9020501138952164,
      "grad_norm": 0.0,
      "learning_rate": 1.8481595092024542e-05,
      "loss": 1.895,
      "step": 396
    },
    {
      "epoch": 0.9043280182232346,
      "grad_norm": 0.0,
      "learning_rate": 1.8475129633186098e-05,
      "loss": 1.7424,
      "step": 397
    },
    {
      "epoch": 0.9066059225512528,
      "grad_norm": 0.0,
      "learning_rate": 1.846864178530204e-05,
      "loss": 1.7934,
      "step": 398
    },
    {
      "epoch": 0.908883826879271,
      "grad_norm": 0.0,
      "learning_rate": 1.8462131431875122e-05,
      "loss": 1.7733,
      "step": 399
    },
    {
      "epoch": 0.9111617312072893,
      "grad_norm": 0.0,
      "learning_rate": 1.8455598455598454e-05,
      "loss": 1.8057,
      "step": 400
    },
    {
      "epoch": 0.9134396355353075,
      "grad_norm": 0.0,
      "learning_rate": 1.8449042738348483e-05,
      "loss": 1.7436,
      "step": 401
    },
    {
      "epoch": 0.9157175398633257,
      "grad_norm": 0.0,
      "learning_rate": 1.844246416117784e-05,
      "loss": 1.842,
      "step": 402
    },
    {
      "epoch": 0.9179954441913439,
      "grad_norm": 0.0,
      "learning_rate": 1.8435862604308172e-05,
      "loss": 1.9916,
      "step": 403
    },
    {
      "epoch": 0.9202733485193622,
      "grad_norm": 0.0,
      "learning_rate": 1.842923794712286e-05,
      "loss": 1.8973,
      "step": 404
    },
    {
      "epoch": 0.9225512528473804,
      "grad_norm": 0.0,
      "learning_rate": 1.842259006815969e-05,
      "loss": 1.8911,
      "step": 405
    },
    {
      "epoch": 0.9248291571753986,
      "grad_norm": 0.0,
      "learning_rate": 1.8415918845103393e-05,
      "loss": 1.8594,
      "step": 406
    },
    {
      "epoch": 0.9271070615034168,
      "grad_norm": 0.0,
      "learning_rate": 1.840922415477819e-05,
      "loss": 1.8265,
      "step": 407
    },
    {
      "epoch": 0.929384965831435,
      "grad_norm": 0.0,
      "learning_rate": 1.8402505873140175e-05,
      "loss": 1.8717,
      "step": 408
    },
    {
      "epoch": 0.9316628701594533,
      "grad_norm": 0.0,
      "learning_rate": 1.8395763875269664e-05,
      "loss": 1.857,
      "step": 409
    },
    {
      "epoch": 0.9339407744874715,
      "grad_norm": 0.0,
      "learning_rate": 1.838899803536346e-05,
      "loss": 1.8825,
      "step": 410
    },
    {
      "epoch": 0.9362186788154897,
      "grad_norm": 0.0,
      "learning_rate": 1.8382208226727024e-05,
      "loss": 1.7305,
      "step": 411
    },
    {
      "epoch": 0.9384965831435079,
      "grad_norm": 0.0,
      "learning_rate": 1.8375394321766563e-05,
      "loss": 1.9017,
      "step": 412
    },
    {
      "epoch": 0.9407744874715261,
      "grad_norm": 0.0,
      "learning_rate": 1.836855619198104e-05,
      "loss": 2.0254,
      "step": 413
    },
    {
      "epoch": 0.9430523917995444,
      "grad_norm": 0.0,
      "learning_rate": 1.83616937079541e-05,
      "loss": 1.9443,
      "step": 414
    },
    {
      "epoch": 0.9453302961275627,
      "grad_norm": 0.0,
      "learning_rate": 1.835480673934589e-05,
      "loss": 1.9315,
      "step": 415
    },
    {
      "epoch": 0.9476082004555809,
      "grad_norm": 0.0,
      "learning_rate": 1.834789515488483e-05,
      "loss": 1.8397,
      "step": 416
    },
    {
      "epoch": 0.9498861047835991,
      "grad_norm": 0.0,
      "learning_rate": 1.834095882235926e-05,
      "loss": 1.6211,
      "step": 417
    },
    {
      "epoch": 0.9521640091116174,
      "grad_norm": 0.0,
      "learning_rate": 1.8333997608609008e-05,
      "loss": 1.8534,
      "step": 418
    },
    {
      "epoch": 0.9544419134396356,
      "grad_norm": 0.0,
      "learning_rate": 1.832701137951687e-05,
      "loss": 1.6394,
      "step": 419
    },
    {
      "epoch": 0.9567198177676538,
      "grad_norm": 0.0,
      "learning_rate": 1.832e-05,
      "loss": 1.7309,
      "step": 420
    },
    {
      "epoch": 0.958997722095672,
      "grad_norm": 0.0,
      "learning_rate": 1.8312963334001203e-05,
      "loss": 1.7431,
      "step": 421
    },
    {
      "epoch": 0.9612756264236902,
      "grad_norm": 0.0,
      "learning_rate": 1.8305901244480132e-05,
      "loss": 1.6638,
      "step": 422
    },
    {
      "epoch": 0.9635535307517085,
      "grad_norm": 0.0,
      "learning_rate": 1.8298813593404385e-05,
      "loss": 1.6408,
      "step": 423
    },
    {
      "epoch": 0.9658314350797267,
      "grad_norm": 0.0,
      "learning_rate": 1.8291700241740533e-05,
      "loss": 1.8247,
      "step": 424
    },
    {
      "epoch": 0.9681093394077449,
      "grad_norm": 0.0,
      "learning_rate": 1.8284561049445004e-05,
      "loss": 1.8894,
      "step": 425
    },
    {
      "epoch": 0.9703872437357631,
      "grad_norm": 0.0,
      "learning_rate": 1.8277395875454913e-05,
      "loss": 1.7693,
      "step": 426
    },
    {
      "epoch": 0.9726651480637813,
      "grad_norm": 0.0,
      "learning_rate": 1.827020457767875e-05,
      "loss": 1.9301,
      "step": 427
    },
    {
      "epoch": 0.9749430523917996,
      "grad_norm": 0.0,
      "learning_rate": 1.8262987012987013e-05,
      "loss": 1.7352,
      "step": 428
    },
    {
      "epoch": 0.9772209567198178,
      "grad_norm": 0.0,
      "learning_rate": 1.8255743037202683e-05,
      "loss": 1.7201,
      "step": 429
    },
    {
      "epoch": 0.979498861047836,
      "grad_norm": 0.0,
      "learning_rate": 1.824847250509165e-05,
      "loss": 1.7524,
      "step": 430
    },
    {
      "epoch": 0.9817767653758542,
      "grad_norm": 0.0,
      "learning_rate": 1.824117527035299e-05,
      "loss": 1.9939,
      "step": 431
    },
    {
      "epoch": 0.9840546697038725,
      "grad_norm": 0.0,
      "learning_rate": 1.823385118560916e-05,
      "loss": 1.7163,
      "step": 432
    },
    {
      "epoch": 0.9863325740318907,
      "grad_norm": 0.0,
      "learning_rate": 1.8226500102396067e-05,
      "loss": 1.8772,
      "step": 433
    },
    {
      "epoch": 0.9886104783599089,
      "grad_norm": 0.0,
      "learning_rate": 1.821912187115306e-05,
      "loss": 1.7732,
      "step": 434
    },
    {
      "epoch": 0.9908883826879271,
      "grad_norm": 0.0,
      "learning_rate": 1.8211716341212746e-05,
      "loss": 1.9645,
      "step": 435
    },
    {
      "epoch": 0.9931662870159453,
      "grad_norm": 0.0,
      "learning_rate": 1.8204283360790775e-05,
      "loss": 1.8126,
      "step": 436
    },
    {
      "epoch": 0.9954441913439636,
      "grad_norm": 0.0,
      "learning_rate": 1.819682277697545e-05,
      "loss": 1.933,
      "step": 437
    },
    {
      "epoch": 0.9977220956719818,
      "grad_norm": 0.0,
      "learning_rate": 1.818933443571724e-05,
      "loss": 2.0602,
      "step": 438
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 2.1612,
      "step": 439
    },
    {
      "epoch": 1.0022779043280183,
      "grad_norm": 0.0,
      "learning_rate": 1.8174273858921162e-05,
      "loss": 1.894,
      "step": 440
    },
    {
      "epoch": 1.0045558086560364,
      "grad_norm": 0.0,
      "learning_rate": 1.8166701309499066e-05,
      "loss": 1.7733,
      "step": 441
    },
    {
      "epoch": 1.0068337129840548,
      "grad_norm": 0.0,
      "learning_rate": 1.8159100374843813e-05,
      "loss": 1.8797,
      "step": 442
    },
    {
      "epoch": 1.0091116173120729,
      "grad_norm": 0.0,
      "learning_rate": 1.815147089505529e-05,
      "loss": 1.851,
      "step": 443
    },
    {
      "epoch": 1.0113895216400912,
      "grad_norm": 0.0,
      "learning_rate": 1.8143812709030104e-05,
      "loss": 1.8259,
      "step": 444
    },
    {
      "epoch": 1.0136674259681093,
      "grad_norm": 0.0,
      "learning_rate": 1.813612565445026e-05,
      "loss": 1.8208,
      "step": 445
    },
    {
      "epoch": 1.0159453302961277,
      "grad_norm": 0.0,
      "learning_rate": 1.812840956777172e-05,
      "loss": 1.8305,
      "step": 446
    },
    {
      "epoch": 1.0182232346241458,
      "grad_norm": 0.0,
      "learning_rate": 1.812066428421274e-05,
      "loss": 1.8653,
      "step": 447
    },
    {
      "epoch": 1.020501138952164,
      "grad_norm": 0.0,
      "learning_rate": 1.811288963774221e-05,
      "loss": 1.6481,
      "step": 448
    },
    {
      "epoch": 1.0227790432801822,
      "grad_norm": 0.0,
      "learning_rate": 1.8105085461067737e-05,
      "loss": 1.9251,
      "step": 449
    },
    {
      "epoch": 1.0250569476082005,
      "grad_norm": 0.0,
      "learning_rate": 1.809725158562368e-05,
      "loss": 1.8392,
      "step": 450
    },
    {
      "epoch": 1.0273348519362187,
      "grad_norm": 0.0,
      "learning_rate": 1.808938784155899e-05,
      "loss": 1.6409,
      "step": 451
    },
    {
      "epoch": 1.029612756264237,
      "grad_norm": 0.0,
      "learning_rate": 1.808149405772496e-05,
      "loss": 1.7448,
      "step": 452
    },
    {
      "epoch": 1.031890660592255,
      "grad_norm": 0.0,
      "learning_rate": 1.807357006166277e-05,
      "loss": 1.687,
      "step": 453
    },
    {
      "epoch": 1.0341685649202734,
      "grad_norm": 0.0,
      "learning_rate": 1.806561567959097e-05,
      "loss": 1.798,
      "step": 454
    },
    {
      "epoch": 1.0364464692482915,
      "grad_norm": 0.0,
      "learning_rate": 1.8057630736392745e-05,
      "loss": 1.9294,
      "step": 455
    },
    {
      "epoch": 1.0387243735763099,
      "grad_norm": 0.0,
      "learning_rate": 1.804961505560308e-05,
      "loss": 1.7994,
      "step": 456
    },
    {
      "epoch": 1.041002277904328,
      "grad_norm": 0.0,
      "learning_rate": 1.804156845939576e-05,
      "loss": 1.8258,
      "step": 457
    },
    {
      "epoch": 1.0432801822323463,
      "grad_norm": 0.0,
      "learning_rate": 1.8033490768570207e-05,
      "loss": 1.7976,
      "step": 458
    },
    {
      "epoch": 1.0455580865603644,
      "grad_norm": 0.0,
      "learning_rate": 1.802538180253818e-05,
      "loss": 1.7135,
      "step": 459
    },
    {
      "epoch": 1.0478359908883828,
      "grad_norm": 0.0,
      "learning_rate": 1.8017241379310346e-05,
      "loss": 1.7839,
      "step": 460
    },
    {
      "epoch": 1.0501138952164009,
      "grad_norm": 0.0,
      "learning_rate": 1.800906931548262e-05,
      "loss": 1.9235,
      "step": 461
    },
    {
      "epoch": 1.0523917995444192,
      "grad_norm": 0.0,
      "learning_rate": 1.8000865426222415e-05,
      "loss": 1.8343,
      "step": 462
    },
    {
      "epoch": 1.0546697038724373,
      "grad_norm": 0.0,
      "learning_rate": 1.7992629525254716e-05,
      "loss": 1.7827,
      "step": 463
    },
    {
      "epoch": 1.0569476082004556,
      "grad_norm": 0.0,
      "learning_rate": 1.798436142484796e-05,
      "loss": 1.6388,
      "step": 464
    },
    {
      "epoch": 1.0592255125284737,
      "grad_norm": 0.0,
      "learning_rate": 1.7976060935799784e-05,
      "loss": 1.8052,
      "step": 465
    },
    {
      "epoch": 1.061503416856492,
      "grad_norm": 0.0,
      "learning_rate": 1.796772786742259e-05,
      "loss": 1.758,
      "step": 466
    },
    {
      "epoch": 1.0637813211845102,
      "grad_norm": 0.0,
      "learning_rate": 1.795936202752895e-05,
      "loss": 1.6495,
      "step": 467
    },
    {
      "epoch": 1.0660592255125285,
      "grad_norm": 0.0,
      "learning_rate": 1.7950963222416815e-05,
      "loss": 1.8546,
      "step": 468
    },
    {
      "epoch": 1.0683371298405466,
      "grad_norm": 0.0,
      "learning_rate": 1.7942531256854574e-05,
      "loss": 1.844,
      "step": 469
    },
    {
      "epoch": 1.070615034168565,
      "grad_norm": 0.0,
      "learning_rate": 1.7934065934065938e-05,
      "loss": 1.8132,
      "step": 470
    },
    {
      "epoch": 1.072892938496583,
      "grad_norm": 0.0,
      "learning_rate": 1.7925567055714605e-05,
      "loss": 1.6304,
      "step": 471
    },
    {
      "epoch": 1.0751708428246014,
      "grad_norm": 0.0,
      "learning_rate": 1.791703442188879e-05,
      "loss": 1.6719,
      "step": 472
    },
    {
      "epoch": 1.0774487471526195,
      "grad_norm": 0.0,
      "learning_rate": 1.7908467831085566e-05,
      "loss": 1.7527,
      "step": 473
    },
    {
      "epoch": 1.0797266514806378,
      "grad_norm": 0.0,
      "learning_rate": 1.789986708019495e-05,
      "loss": 1.8619,
      "step": 474
    },
    {
      "epoch": 1.082004555808656,
      "grad_norm": 0.0,
      "learning_rate": 1.7891231964483905e-05,
      "loss": 1.759,
      "step": 475
    },
    {
      "epoch": 1.0842824601366743,
      "grad_norm": 0.0,
      "learning_rate": 1.7882562277580073e-05,
      "loss": 1.8707,
      "step": 476
    },
    {
      "epoch": 1.0865603644646924,
      "grad_norm": 0.0,
      "learning_rate": 1.7873857811455315e-05,
      "loss": 1.8425,
      "step": 477
    },
    {
      "epoch": 1.0888382687927107,
      "grad_norm": 0.0,
      "learning_rate": 1.786511835640911e-05,
      "loss": 1.862,
      "step": 478
    },
    {
      "epoch": 1.0911161731207288,
      "grad_norm": 0.0,
      "learning_rate": 1.785634370105169e-05,
      "loss": 1.7945,
      "step": 479
    },
    {
      "epoch": 1.0933940774487472,
      "grad_norm": 0.0,
      "learning_rate": 1.7847533632286997e-05,
      "loss": 1.775,
      "step": 480
    },
    {
      "epoch": 1.0956719817767653,
      "grad_norm": 0.0,
      "learning_rate": 1.783868793529544e-05,
      "loss": 1.9576,
      "step": 481
    },
    {
      "epoch": 1.0979498861047836,
      "grad_norm": 0.0,
      "learning_rate": 1.7829806393516437e-05,
      "loss": 1.6999,
      "step": 482
    },
    {
      "epoch": 1.1002277904328017,
      "grad_norm": 0.0,
      "learning_rate": 1.7820888788630725e-05,
      "loss": 1.9236,
      "step": 483
    },
    {
      "epoch": 1.10250569476082,
      "grad_norm": 0.0,
      "learning_rate": 1.7811934900542496e-05,
      "loss": 1.8691,
      "step": 484
    },
    {
      "epoch": 1.1047835990888384,
      "grad_norm": 0.0,
      "learning_rate": 1.780294450736127e-05,
      "loss": 1.7682,
      "step": 485
    },
    {
      "epoch": 1.1070615034168565,
      "grad_norm": 0.0,
      "learning_rate": 1.7793917385383568e-05,
      "loss": 1.7202,
      "step": 486
    },
    {
      "epoch": 1.1093394077448746,
      "grad_norm": 0.0,
      "learning_rate": 1.7784853309074372e-05,
      "loss": 1.9333,
      "step": 487
    },
    {
      "epoch": 1.111617312072893,
      "grad_norm": 0.0,
      "learning_rate": 1.7775752051048316e-05,
      "loss": 1.6755,
      "step": 488
    },
    {
      "epoch": 1.1138952164009113,
      "grad_norm": 0.0,
      "learning_rate": 1.7766613382050697e-05,
      "loss": 1.7697,
      "step": 489
    },
    {
      "epoch": 1.1161731207289294,
      "grad_norm": 0.0,
      "learning_rate": 1.7757437070938218e-05,
      "loss": 1.7659,
      "step": 490
    },
    {
      "epoch": 1.1184510250569477,
      "grad_norm": 0.0,
      "learning_rate": 1.7748222884659486e-05,
      "loss": 2.0891,
      "step": 491
    },
    {
      "epoch": 1.1207289293849658,
      "grad_norm": 0.0,
      "learning_rate": 1.7738970588235293e-05,
      "loss": 1.7104,
      "step": 492
    },
    {
      "epoch": 1.1230068337129842,
      "grad_norm": 0.0,
      "learning_rate": 1.7729679944738662e-05,
      "loss": 1.9044,
      "step": 493
    },
    {
      "epoch": 1.1252847380410023,
      "grad_norm": 0.0,
      "learning_rate": 1.7720350715274574e-05,
      "loss": 1.8321,
      "step": 494
    },
    {
      "epoch": 1.1275626423690206,
      "grad_norm": 0.0,
      "learning_rate": 1.771098265895954e-05,
      "loss": 1.7164,
      "step": 495
    },
    {
      "epoch": 1.1298405466970387,
      "grad_norm": 0.0,
      "learning_rate": 1.7701575532900836e-05,
      "loss": 1.9128,
      "step": 496
    },
    {
      "epoch": 1.132118451025057,
      "grad_norm": 0.0,
      "learning_rate": 1.7692129092175527e-05,
      "loss": 1.7434,
      "step": 497
    },
    {
      "epoch": 1.1343963553530751,
      "grad_norm": 0.0,
      "learning_rate": 1.7682643089809216e-05,
      "loss": 1.7226,
      "step": 498
    },
    {
      "epoch": 1.1366742596810935,
      "grad_norm": 0.0,
      "learning_rate": 1.767311727675449e-05,
      "loss": 1.7137,
      "step": 499
    },
    {
      "epoch": 1.1389521640091116,
      "grad_norm": 0.0,
      "learning_rate": 1.7663551401869158e-05,
      "loss": 1.8651,
      "step": 500
    },
    {
      "epoch": 1.14123006833713,
      "grad_norm": 0.0,
      "learning_rate": 1.765394521189417e-05,
      "loss": 1.8598,
      "step": 501
    },
    {
      "epoch": 1.143507972665148,
      "grad_norm": 0.0,
      "learning_rate": 1.7644298451431255e-05,
      "loss": 1.7573,
      "step": 502
    },
    {
      "epoch": 1.1457858769931664,
      "grad_norm": 0.0,
      "learning_rate": 1.7634610862920293e-05,
      "loss": 1.9351,
      "step": 503
    },
    {
      "epoch": 1.1480637813211845,
      "grad_norm": 0.0,
      "learning_rate": 1.76248821866164e-05,
      "loss": 1.9187,
      "step": 504
    },
    {
      "epoch": 1.1503416856492028,
      "grad_norm": 0.0,
      "learning_rate": 1.761511216056671e-05,
      "loss": 1.8173,
      "step": 505
    },
    {
      "epoch": 1.152619589977221,
      "grad_norm": 0.0,
      "learning_rate": 1.7605300520586845e-05,
      "loss": 1.6676,
      "step": 506
    },
    {
      "epoch": 1.1548974943052392,
      "grad_norm": 0.0,
      "learning_rate": 1.7595447000237135e-05,
      "loss": 1.6559,
      "step": 507
    },
    {
      "epoch": 1.1571753986332574,
      "grad_norm": 0.0,
      "learning_rate": 1.758555133079848e-05,
      "loss": 1.6879,
      "step": 508
    },
    {
      "epoch": 1.1594533029612757,
      "grad_norm": 0.0,
      "learning_rate": 1.7575613241247916e-05,
      "loss": 1.8797,
      "step": 509
    },
    {
      "epoch": 1.1617312072892938,
      "grad_norm": 0.0,
      "learning_rate": 1.7565632458233892e-05,
      "loss": 1.8194,
      "step": 510
    },
    {
      "epoch": 1.1640091116173121,
      "grad_norm": 0.0,
      "learning_rate": 1.7555608706051185e-05,
      "loss": 1.7954,
      "step": 511
    },
    {
      "epoch": 1.1662870159453302,
      "grad_norm": 0.0,
      "learning_rate": 1.754554170661553e-05,
      "loss": 1.797,
      "step": 512
    },
    {
      "epoch": 1.1685649202733486,
      "grad_norm": 0.0,
      "learning_rate": 1.7535431179437905e-05,
      "loss": 1.9259,
      "step": 513
    },
    {
      "epoch": 1.1708428246013667,
      "grad_norm": 0.0,
      "learning_rate": 1.752527684159846e-05,
      "loss": 1.8345,
      "step": 514
    },
    {
      "epoch": 1.173120728929385,
      "grad_norm": 0.0,
      "learning_rate": 1.7515078407720147e-05,
      "loss": 1.771,
      "step": 515
    },
    {
      "epoch": 1.1753986332574031,
      "grad_norm": 0.0,
      "learning_rate": 1.7504835589941973e-05,
      "loss": 1.8106,
      "step": 516
    },
    {
      "epoch": 1.1776765375854215,
      "grad_norm": 0.0,
      "learning_rate": 1.749454809789193e-05,
      "loss": 1.978,
      "step": 517
    },
    {
      "epoch": 1.1799544419134396,
      "grad_norm": 0.0,
      "learning_rate": 1.7484215638659544e-05,
      "loss": 1.9003,
      "step": 518
    },
    {
      "epoch": 1.182232346241458,
      "grad_norm": 0.0,
      "learning_rate": 1.7473837916768072e-05,
      "loss": 1.7944,
      "step": 519
    },
    {
      "epoch": 1.184510250569476,
      "grad_norm": 0.0,
      "learning_rate": 1.7463414634146342e-05,
      "loss": 1.793,
      "step": 520
    },
    {
      "epoch": 1.1867881548974943,
      "grad_norm": 0.0,
      "learning_rate": 1.745294549010022e-05,
      "loss": 1.8594,
      "step": 521
    },
    {
      "epoch": 1.1890660592255125,
      "grad_norm": 0.0,
      "learning_rate": 1.7442430181283685e-05,
      "loss": 1.7914,
      "step": 522
    },
    {
      "epoch": 1.1913439635535308,
      "grad_norm": 0.0,
      "learning_rate": 1.7431868401669533e-05,
      "loss": 1.8498,
      "step": 523
    },
    {
      "epoch": 1.193621867881549,
      "grad_norm": 0.0,
      "learning_rate": 1.7421259842519685e-05,
      "loss": 1.9455,
      "step": 524
    },
    {
      "epoch": 1.1958997722095672,
      "grad_norm": 0.0,
      "learning_rate": 1.7410604192355117e-05,
      "loss": 1.7882,
      "step": 525
    },
    {
      "epoch": 1.1981776765375853,
      "grad_norm": 0.0,
      "learning_rate": 1.739990113692536e-05,
      "loss": 1.8283,
      "step": 526
    },
    {
      "epoch": 1.2004555808656037,
      "grad_norm": 0.0,
      "learning_rate": 1.7389150359177605e-05,
      "loss": 1.6758,
      "step": 527
    },
    {
      "epoch": 1.2027334851936218,
      "grad_norm": 0.0,
      "learning_rate": 1.7378351539225422e-05,
      "loss": 1.7471,
      "step": 528
    },
    {
      "epoch": 1.20501138952164,
      "grad_norm": 0.0,
      "learning_rate": 1.7367504354316996e-05,
      "loss": 1.7923,
      "step": 529
    },
    {
      "epoch": 1.2072892938496582,
      "grad_norm": 0.0,
      "learning_rate": 1.7356608478802993e-05,
      "loss": 1.8953,
      "step": 530
    },
    {
      "epoch": 1.2095671981776766,
      "grad_norm": 0.0,
      "learning_rate": 1.7345663584103974e-05,
      "loss": 1.8777,
      "step": 531
    },
    {
      "epoch": 1.2118451025056949,
      "grad_norm": 0.0,
      "learning_rate": 1.7334669338677356e-05,
      "loss": 1.8428,
      "step": 532
    },
    {
      "epoch": 1.214123006833713,
      "grad_norm": 0.0,
      "learning_rate": 1.7323625407983932e-05,
      "loss": 1.5904,
      "step": 533
    },
    {
      "epoch": 1.216400911161731,
      "grad_norm": 0.0,
      "learning_rate": 1.731253145445395e-05,
      "loss": 1.6985,
      "step": 534
    },
    {
      "epoch": 1.2186788154897494,
      "grad_norm": 0.0,
      "learning_rate": 1.7301387137452713e-05,
      "loss": 1.6751,
      "step": 535
    },
    {
      "epoch": 1.2209567198177678,
      "grad_norm": 0.0,
      "learning_rate": 1.7290192113245704e-05,
      "loss": 1.7691,
      "step": 536
    },
    {
      "epoch": 1.2232346241457859,
      "grad_norm": 0.0,
      "learning_rate": 1.7278946034963264e-05,
      "loss": 1.8451,
      "step": 537
    },
    {
      "epoch": 1.225512528473804,
      "grad_norm": 0.0,
      "learning_rate": 1.7267648552564757e-05,
      "loss": 1.7276,
      "step": 538
    },
    {
      "epoch": 1.2277904328018223,
      "grad_norm": 0.0,
      "learning_rate": 1.725629931280224e-05,
      "loss": 1.7447,
      "step": 539
    },
    {
      "epoch": 1.2300683371298406,
      "grad_norm": 0.0,
      "learning_rate": 1.7244897959183674e-05,
      "loss": 1.8618,
      "step": 540
    },
    {
      "epoch": 1.2323462414578588,
      "grad_norm": 0.0,
      "learning_rate": 1.7233444131935567e-05,
      "loss": 1.909,
      "step": 541
    },
    {
      "epoch": 1.2346241457858769,
      "grad_norm": 0.0,
      "learning_rate": 1.7221937467965147e-05,
      "loss": 1.7861,
      "step": 542
    },
    {
      "epoch": 1.2369020501138952,
      "grad_norm": 0.0,
      "learning_rate": 1.7210377600821992e-05,
      "loss": 1.8093,
      "step": 543
    },
    {
      "epoch": 1.2391799544419135,
      "grad_norm": 0.0,
      "learning_rate": 1.7198764160659114e-05,
      "loss": 1.8772,
      "step": 544
    },
    {
      "epoch": 1.2414578587699316,
      "grad_norm": 0.0,
      "learning_rate": 1.7187096774193548e-05,
      "loss": 1.7474,
      "step": 545
    },
    {
      "epoch": 1.24373576309795,
      "grad_norm": 0.0,
      "learning_rate": 1.7175375064666324e-05,
      "loss": 1.8976,
      "step": 546
    },
    {
      "epoch": 1.246013667425968,
      "grad_norm": 0.0,
      "learning_rate": 1.716359865180192e-05,
      "loss": 1.9038,
      "step": 547
    },
    {
      "epoch": 1.2482915717539864,
      "grad_norm": 0.0,
      "learning_rate": 1.7151767151767152e-05,
      "loss": 1.8241,
      "step": 548
    },
    {
      "epoch": 1.2505694760820045,
      "grad_norm": 0.0,
      "learning_rate": 1.7139880177129463e-05,
      "loss": 1.8321,
      "step": 549
    },
    {
      "epoch": 1.2528473804100229,
      "grad_norm": 0.0,
      "learning_rate": 1.7127937336814622e-05,
      "loss": 1.7573,
      "step": 550
    },
    {
      "epoch": 1.255125284738041,
      "grad_norm": 0.0,
      "learning_rate": 1.7115938236063858e-05,
      "loss": 1.8974,
      "step": 551
    },
    {
      "epoch": 1.2574031890660593,
      "grad_norm": 0.0,
      "learning_rate": 1.710388247639035e-05,
      "loss": 1.6982,
      "step": 552
    },
    {
      "epoch": 1.2596810933940774,
      "grad_norm": 0.0,
      "learning_rate": 1.7091769655535105e-05,
      "loss": 1.8387,
      "step": 553
    },
    {
      "epoch": 1.2619589977220957,
      "grad_norm": 0.0,
      "learning_rate": 1.7079599367422247e-05,
      "loss": 1.935,
      "step": 554
    },
    {
      "epoch": 1.2642369020501139,
      "grad_norm": 0.0,
      "learning_rate": 1.706737120211361e-05,
      "loss": 1.873,
      "step": 555
    },
    {
      "epoch": 1.2665148063781322,
      "grad_norm": 0.0,
      "learning_rate": 1.705508474576271e-05,
      "loss": 1.9897,
      "step": 556
    },
    {
      "epoch": 1.2687927107061503,
      "grad_norm": 0.0,
      "learning_rate": 1.7042739580568093e-05,
      "loss": 1.9435,
      "step": 557
    },
    {
      "epoch": 1.2710706150341686,
      "grad_norm": 0.0,
      "learning_rate": 1.703033528472592e-05,
      "loss": 2.0429,
      "step": 558
    },
    {
      "epoch": 1.2733485193621867,
      "grad_norm": 0.0,
      "learning_rate": 1.701787143238197e-05,
      "loss": 1.7305,
      "step": 559
    },
    {
      "epoch": 1.275626423690205,
      "grad_norm": 0.0,
      "learning_rate": 1.700534759358289e-05,
      "loss": 1.8407,
      "step": 560
    },
    {
      "epoch": 1.2779043280182232,
      "grad_norm": 0.0,
      "learning_rate": 1.699276333422675e-05,
      "loss": 1.8374,
      "step": 561
    },
    {
      "epoch": 1.2801822323462415,
      "grad_norm": 0.0,
      "learning_rate": 1.69801182160129e-05,
      "loss": 1.8984,
      "step": 562
    },
    {
      "epoch": 1.2824601366742596,
      "grad_norm": 0.0,
      "learning_rate": 1.696741179639106e-05,
      "loss": 1.8733,
      "step": 563
    },
    {
      "epoch": 1.284738041002278,
      "grad_norm": 0.0,
      "learning_rate": 1.695464362850972e-05,
      "loss": 1.7977,
      "step": 564
    },
    {
      "epoch": 1.287015945330296,
      "grad_norm": 0.0,
      "learning_rate": 1.6941813261163737e-05,
      "loss": 1.8713,
      "step": 565
    },
    {
      "epoch": 1.2892938496583144,
      "grad_norm": 0.0,
      "learning_rate": 1.692892023874118e-05,
      "loss": 2.0701,
      "step": 566
    },
    {
      "epoch": 1.2915717539863325,
      "grad_norm": 0.0,
      "learning_rate": 1.6915964101169432e-05,
      "loss": 1.9373,
      "step": 567
    },
    {
      "epoch": 1.2938496583143508,
      "grad_norm": 0.0,
      "learning_rate": 1.6902944383860415e-05,
      "loss": 1.9324,
      "step": 568
    },
    {
      "epoch": 1.296127562642369,
      "grad_norm": 0.0,
      "learning_rate": 1.6889860617655097e-05,
      "loss": 1.8836,
      "step": 569
    },
    {
      "epoch": 1.2984054669703873,
      "grad_norm": 0.0,
      "learning_rate": 1.6876712328767124e-05,
      "loss": 1.6978,
      "step": 570
    },
    {
      "epoch": 1.3006833712984054,
      "grad_norm": 0.0,
      "learning_rate": 1.6863499038725624e-05,
      "loss": 1.7005,
      "step": 571
    },
    {
      "epoch": 1.3029612756264237,
      "grad_norm": 0.0,
      "learning_rate": 1.685022026431718e-05,
      "loss": 1.7452,
      "step": 572
    },
    {
      "epoch": 1.3052391799544418,
      "grad_norm": 0.0,
      "learning_rate": 1.6836875517526915e-05,
      "loss": 1.7374,
      "step": 573
    },
    {
      "epoch": 1.3075170842824602,
      "grad_norm": 0.0,
      "learning_rate": 1.6823464305478693e-05,
      "loss": 1.955,
      "step": 574
    },
    {
      "epoch": 1.3097949886104785,
      "grad_norm": 0.0,
      "learning_rate": 1.680998613037448e-05,
      "loss": 1.9013,
      "step": 575
    },
    {
      "epoch": 1.3120728929384966,
      "grad_norm": 0.0,
      "learning_rate": 1.6796440489432706e-05,
      "loss": 1.771,
      "step": 576
    },
    {
      "epoch": 1.3143507972665147,
      "grad_norm": 0.0,
      "learning_rate": 1.678282687482576e-05,
      "loss": 1.8485,
      "step": 577
    },
    {
      "epoch": 1.316628701594533,
      "grad_norm": 0.0,
      "learning_rate": 1.6769144773616545e-05,
      "loss": 1.8416,
      "step": 578
    },
    {
      "epoch": 1.3189066059225514,
      "grad_norm": 0.0,
      "learning_rate": 1.6755393667694036e-05,
      "loss": 1.83,
      "step": 579
    },
    {
      "epoch": 1.3211845102505695,
      "grad_norm": 0.0,
      "learning_rate": 1.6741573033707864e-05,
      "loss": 1.7065,
      "step": 580
    },
    {
      "epoch": 1.3234624145785876,
      "grad_norm": 0.0,
      "learning_rate": 1.6727682343001974e-05,
      "loss": 1.72,
      "step": 581
    },
    {
      "epoch": 1.325740318906606,
      "grad_norm": 0.0,
      "learning_rate": 1.6713721061547152e-05,
      "loss": 1.7954,
      "step": 582
    },
    {
      "epoch": 1.3280182232346243,
      "grad_norm": 0.0,
      "learning_rate": 1.6699688649872628e-05,
      "loss": 1.7777,
      "step": 583
    },
    {
      "epoch": 1.3302961275626424,
      "grad_norm": 0.0,
      "learning_rate": 1.6685584562996594e-05,
      "loss": 1.6499,
      "step": 584
    },
    {
      "epoch": 1.3325740318906605,
      "grad_norm": 0.0,
      "learning_rate": 1.667140825035562e-05,
      "loss": 1.9445,
      "step": 585
    },
    {
      "epoch": 1.3348519362186788,
      "grad_norm": 0.0,
      "learning_rate": 1.665715915573303e-05,
      "loss": 1.7321,
      "step": 586
    },
    {
      "epoch": 1.3371298405466971,
      "grad_norm": 0.0,
      "learning_rate": 1.6642836717186162e-05,
      "loss": 1.8359,
      "step": 587
    },
    {
      "epoch": 1.3394077448747153,
      "grad_norm": 0.0,
      "learning_rate": 1.6628440366972475e-05,
      "loss": 1.8457,
      "step": 588
    },
    {
      "epoch": 1.3416856492027334,
      "grad_norm": 0.0,
      "learning_rate": 1.6613969531474562e-05,
      "loss": 1.6885,
      "step": 589
    },
    {
      "epoch": 1.3439635535307517,
      "grad_norm": 0.0,
      "learning_rate": 1.6599423631123922e-05,
      "loss": 1.8215,
      "step": 590
    },
    {
      "epoch": 1.34624145785877,
      "grad_norm": 0.0,
      "learning_rate": 1.6584802080323605e-05,
      "loss": 2.0192,
      "step": 591
    },
    {
      "epoch": 1.3485193621867881,
      "grad_norm": 0.0,
      "learning_rate": 1.6570104287369643e-05,
      "loss": 1.7816,
      "step": 592
    },
    {
      "epoch": 1.3507972665148062,
      "grad_norm": 0.0,
      "learning_rate": 1.655532965437119e-05,
      "loss": 1.8412,
      "step": 593
    },
    {
      "epoch": 1.3530751708428246,
      "grad_norm": 0.0,
      "learning_rate": 1.654047757716948e-05,
      "loss": 1.7873,
      "step": 594
    },
    {
      "epoch": 1.355353075170843,
      "grad_norm": 0.0,
      "learning_rate": 1.6525547445255474e-05,
      "loss": 1.6725,
      "step": 595
    },
    {
      "epoch": 1.357630979498861,
      "grad_norm": 0.0,
      "learning_rate": 1.6510538641686185e-05,
      "loss": 1.8629,
      "step": 596
    },
    {
      "epoch": 1.3599088838268791,
      "grad_norm": 0.0,
      "learning_rate": 1.649545054299971e-05,
      "loss": 1.7792,
      "step": 597
    },
    {
      "epoch": 1.3621867881548975,
      "grad_norm": 0.0,
      "learning_rate": 1.64802825191289e-05,
      "loss": 1.7525,
      "step": 598
    },
    {
      "epoch": 1.3644646924829158,
      "grad_norm": 0.0,
      "learning_rate": 1.6465033933313664e-05,
      "loss": 1.7946,
      "step": 599
    },
    {
      "epoch": 1.366742596810934,
      "grad_norm": 0.0,
      "learning_rate": 1.6449704142011833e-05,
      "loss": 1.8881,
      "step": 600
    },
    {
      "epoch": 1.3690205011389522,
      "grad_norm": 0.0,
      "learning_rate": 1.6434292494808663e-05,
      "loss": 1.7876,
      "step": 601
    },
    {
      "epoch": 1.3712984054669703,
      "grad_norm": 0.0,
      "learning_rate": 1.641879833432481e-05,
      "loss": 1.726,
      "step": 602
    },
    {
      "epoch": 1.3735763097949887,
      "grad_norm": 0.0,
      "learning_rate": 1.6403220996122876e-05,
      "loss": 1.8706,
      "step": 603
    },
    {
      "epoch": 1.3758542141230068,
      "grad_norm": 0.0,
      "learning_rate": 1.638755980861244e-05,
      "loss": 1.7448,
      "step": 604
    },
    {
      "epoch": 1.3781321184510251,
      "grad_norm": 0.0,
      "learning_rate": 1.6371814092953527e-05,
      "loss": 1.94,
      "step": 605
    },
    {
      "epoch": 1.3804100227790432,
      "grad_norm": 0.0,
      "learning_rate": 1.6355983162958508e-05,
      "loss": 1.8216,
      "step": 606
    },
    {
      "epoch": 1.3826879271070616,
      "grad_norm": 0.0,
      "learning_rate": 1.6340066324992464e-05,
      "loss": 1.8271,
      "step": 607
    },
    {
      "epoch": 1.3849658314350797,
      "grad_norm": 0.0,
      "learning_rate": 1.6324062877871826e-05,
      "loss": 1.9633,
      "step": 608
    },
    {
      "epoch": 1.387243735763098,
      "grad_norm": 0.0,
      "learning_rate": 1.6307972112761445e-05,
      "loss": 1.8003,
      "step": 609
    },
    {
      "epoch": 1.3895216400911161,
      "grad_norm": 0.0,
      "learning_rate": 1.629179331306991e-05,
      "loss": 1.7099,
      "step": 610
    },
    {
      "epoch": 1.3917995444191344,
      "grad_norm": 0.0,
      "learning_rate": 1.6275525754343188e-05,
      "loss": 1.8283,
      "step": 611
    },
    {
      "epoch": 1.3940774487471526,
      "grad_norm": 0.0,
      "learning_rate": 1.6259168704156484e-05,
      "loss": 1.6879,
      "step": 612
    },
    {
      "epoch": 1.396355353075171,
      "grad_norm": 0.0,
      "learning_rate": 1.624272142200429e-05,
      "loss": 1.8087,
      "step": 613
    },
    {
      "epoch": 1.398633257403189,
      "grad_norm": 0.0,
      "learning_rate": 1.6226183159188693e-05,
      "loss": 1.6436,
      "step": 614
    },
    {
      "epoch": 1.4009111617312073,
      "grad_norm": 0.0,
      "learning_rate": 1.62095531587057e-05,
      "loss": 1.8455,
      "step": 615
    },
    {
      "epoch": 1.4031890660592254,
      "grad_norm": 0.0,
      "learning_rate": 1.619283065512979e-05,
      "loss": 1.7757,
      "step": 616
    },
    {
      "epoch": 1.4054669703872438,
      "grad_norm": 0.0,
      "learning_rate": 1.6176014874496436e-05,
      "loss": 1.7478,
      "step": 617
    },
    {
      "epoch": 1.4077448747152619,
      "grad_norm": 0.0,
      "learning_rate": 1.615910503418272e-05,
      "loss": 1.7987,
      "step": 618
    },
    {
      "epoch": 1.4100227790432802,
      "grad_norm": 0.0,
      "learning_rate": 1.6142100342785915e-05,
      "loss": 1.8381,
      "step": 619
    },
    {
      "epoch": 1.4123006833712983,
      "grad_norm": 0.0,
      "learning_rate": 1.6125000000000002e-05,
      "loss": 1.8054,
      "step": 620
    },
    {
      "epoch": 1.4145785876993167,
      "grad_norm": 0.0,
      "learning_rate": 1.6107803196490126e-05,
      "loss": 1.878,
      "step": 621
    },
    {
      "epoch": 1.416856492027335,
      "grad_norm": 0.0,
      "learning_rate": 1.6090509113764927e-05,
      "loss": 1.8996,
      "step": 622
    },
    {
      "epoch": 1.419134396355353,
      "grad_norm": 0.0,
      "learning_rate": 1.6073116924046647e-05,
      "loss": 1.6438,
      "step": 623
    },
    {
      "epoch": 1.4214123006833712,
      "grad_norm": 0.0,
      "learning_rate": 1.6055625790139063e-05,
      "loss": 1.9268,
      "step": 624
    },
    {
      "epoch": 1.4236902050113895,
      "grad_norm": 0.0,
      "learning_rate": 1.6038034865293186e-05,
      "loss": 1.7209,
      "step": 625
    },
    {
      "epoch": 1.4259681093394079,
      "grad_norm": 0.0,
      "learning_rate": 1.6020343293070568e-05,
      "loss": 1.7906,
      "step": 626
    },
    {
      "epoch": 1.428246013667426,
      "grad_norm": 0.0,
      "learning_rate": 1.6002550207204335e-05,
      "loss": 1.779,
      "step": 627
    },
    {
      "epoch": 1.430523917995444,
      "grad_norm": 0.0,
      "learning_rate": 1.59846547314578e-05,
      "loss": 1.6729,
      "step": 628
    },
    {
      "epoch": 1.4328018223234624,
      "grad_norm": 0.0,
      "learning_rate": 1.5966655979480607e-05,
      "loss": 1.7129,
      "step": 629
    },
    {
      "epoch": 1.4350797266514808,
      "grad_norm": 0.0,
      "learning_rate": 1.594855305466238e-05,
      "loss": 1.7471,
      "step": 630
    },
    {
      "epoch": 1.4373576309794989,
      "grad_norm": 0.0,
      "learning_rate": 1.5930345049983877e-05,
      "loss": 1.7756,
      "step": 631
    },
    {
      "epoch": 1.439635535307517,
      "grad_norm": 0.0,
      "learning_rate": 1.591203104786546e-05,
      "loss": 1.8302,
      "step": 632
    },
    {
      "epoch": 1.4419134396355353,
      "grad_norm": 0.0,
      "learning_rate": 1.5893610120012975e-05,
      "loss": 1.7776,
      "step": 633
    },
    {
      "epoch": 1.4441913439635536,
      "grad_norm": 0.0,
      "learning_rate": 1.58750813272609e-05,
      "loss": 1.7861,
      "step": 634
    },
    {
      "epoch": 1.4464692482915718,
      "grad_norm": 0.0,
      "learning_rate": 1.5856443719412723e-05,
      "loss": 1.8694,
      "step": 635
    },
    {
      "epoch": 1.4487471526195899,
      "grad_norm": 0.0,
      "learning_rate": 1.5837696335078534e-05,
      "loss": 1.7036,
      "step": 636
    },
    {
      "epoch": 1.4510250569476082,
      "grad_norm": 0.0,
      "learning_rate": 1.5818838201509683e-05,
      "loss": 1.9997,
      "step": 637
    },
    {
      "epoch": 1.4533029612756265,
      "grad_norm": 0.0,
      "learning_rate": 1.579986833443055e-05,
      "loss": 1.8658,
      "step": 638
    },
    {
      "epoch": 1.4555808656036446,
      "grad_norm": 0.0,
      "learning_rate": 1.5780785737867282e-05,
      "loss": 1.7876,
      "step": 639
    },
    {
      "epoch": 1.4578587699316627,
      "grad_norm": 0.0,
      "learning_rate": 1.576158940397351e-05,
      "loss": 1.8197,
      "step": 640
    },
    {
      "epoch": 1.460136674259681,
      "grad_norm": 0.0,
      "learning_rate": 1.5742278312852876e-05,
      "loss": 1.7887,
      "step": 641
    },
    {
      "epoch": 1.4624145785876994,
      "grad_norm": 0.0,
      "learning_rate": 1.5722851432378414e-05,
      "loss": 2.0391,
      "step": 642
    },
    {
      "epoch": 1.4646924829157175,
      "grad_norm": 0.0,
      "learning_rate": 1.570330771800869e-05,
      "loss": 1.9192,
      "step": 643
    },
    {
      "epoch": 1.4669703872437356,
      "grad_norm": 0.0,
      "learning_rate": 1.5683646112600534e-05,
      "loss": 1.7891,
      "step": 644
    },
    {
      "epoch": 1.469248291571754,
      "grad_norm": 0.0,
      "learning_rate": 1.5663865546218488e-05,
      "loss": 1.7657,
      "step": 645
    },
    {
      "epoch": 1.4715261958997723,
      "grad_norm": 0.0,
      "learning_rate": 1.5643964935940662e-05,
      "loss": 1.8506,
      "step": 646
    },
    {
      "epoch": 1.4738041002277904,
      "grad_norm": 0.0,
      "learning_rate": 1.5623943185661144e-05,
      "loss": 1.8685,
      "step": 647
    },
    {
      "epoch": 1.4760820045558087,
      "grad_norm": 0.0,
      "learning_rate": 1.5603799185888738e-05,
      "loss": 1.7975,
      "step": 648
    },
    {
      "epoch": 1.4783599088838268,
      "grad_norm": 0.0,
      "learning_rate": 1.5583531813542022e-05,
      "loss": 1.8862,
      "step": 649
    },
    {
      "epoch": 1.4806378132118452,
      "grad_norm": 0.0,
      "learning_rate": 1.5563139931740613e-05,
      "loss": 1.7348,
      "step": 650
    },
    {
      "epoch": 1.4829157175398633,
      "grad_norm": 0.0,
      "learning_rate": 1.5542622389592607e-05,
      "loss": 1.6889,
      "step": 651
    },
    {
      "epoch": 1.4851936218678816,
      "grad_norm": 0.0,
      "learning_rate": 1.5521978021978023e-05,
      "loss": 1.9386,
      "step": 652
    },
    {
      "epoch": 1.4874715261958997,
      "grad_norm": 0.0,
      "learning_rate": 1.550120564932828e-05,
      "loss": 1.9238,
      "step": 653
    },
    {
      "epoch": 1.489749430523918,
      "grad_norm": 0.0,
      "learning_rate": 1.548030407740152e-05,
      "loss": 1.8025,
      "step": 654
    },
    {
      "epoch": 1.4920273348519362,
      "grad_norm": 0.0,
      "learning_rate": 1.545927209705373e-05,
      "loss": 1.7167,
      "step": 655
    },
    {
      "epoch": 1.4943052391799545,
      "grad_norm": 0.0,
      "learning_rate": 1.5438108484005562e-05,
      "loss": 1.7308,
      "step": 656
    },
    {
      "epoch": 1.4965831435079726,
      "grad_norm": 0.0,
      "learning_rate": 1.5416811998604815e-05,
      "loss": 1.8564,
      "step": 657
    },
    {
      "epoch": 1.498861047835991,
      "grad_norm": 0.0,
      "learning_rate": 1.5395381385584325e-05,
      "loss": 1.621,
      "step": 658
    },
    {
      "epoch": 1.501138952164009,
      "grad_norm": 0.0,
      "learning_rate": 1.537381537381537e-05,
      "loss": 1.8461,
      "step": 659
    },
    {
      "epoch": 1.5034168564920274,
      "grad_norm": 0.0,
      "learning_rate": 1.535211267605634e-05,
      "loss": 1.859,
      "step": 660
    }
  ],
  "logging_steps": 1,
  "max_steps": 878,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 220,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 7.847387829381366e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}