{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 103194,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.975919142585809e-05,
      "loss": 2.0152,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9517413803128096e-05,
      "loss": 1.7371,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.927515165610404e-05,
      "loss": 1.6658,
      "step": 1500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.903288950907999e-05,
      "loss": 1.5673,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.879062736205593e-05,
      "loss": 1.5484,
      "step": 2500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8548365215031885e-05,
      "loss": 1.4843,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.830610306800783e-05,
      "loss": 1.4821,
      "step": 3500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8063840920983784e-05,
      "loss": 1.4424,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.782157877395973e-05,
      "loss": 1.4388,
      "step": 4500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7579316626935675e-05,
      "loss": 1.3804,
      "step": 5000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.733705447991163e-05,
      "loss": 1.402,
      "step": 5500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7094792332887573e-05,
      "loss": 1.3506,
      "step": 6000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.685301471015757e-05,
      "loss": 1.3347,
      "step": 6500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.661075256313352e-05,
      "loss": 1.3199,
      "step": 7000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.636849041610947e-05,
      "loss": 1.3252,
      "step": 7500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6126228269085415e-05,
      "loss": 1.2724,
      "step": 8000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.588396612206136e-05,
      "loss": 1.3366,
      "step": 8500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.564170397503731e-05,
      "loss": 1.2848,
      "step": 9000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.539944182801326e-05,
      "loss": 1.2723,
      "step": 9500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.515717968098921e-05,
      "loss": 1.299,
      "step": 10000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.491491753396516e-05,
      "loss": 1.3005,
      "step": 10500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.46726553869411e-05,
      "loss": 1.2675,
      "step": 11000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.443039323991705e-05,
      "loss": 1.2606,
      "step": 11500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4188131092892995e-05,
      "loss": 1.2592,
      "step": 12000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.3946353470163e-05,
      "loss": 1.2159,
      "step": 12500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.3704091323138945e-05,
      "loss": 1.213,
      "step": 13000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.346182917611489e-05,
      "loss": 1.2297,
      "step": 13500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.3219567029090844e-05,
      "loss": 1.2048,
      "step": 14000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.2977789406360835e-05,
      "loss": 1.2324,
      "step": 14500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.273552725933678e-05,
      "loss": 1.2067,
      "step": 15000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.249326511231273e-05,
      "loss": 1.2146,
      "step": 15500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2251002965288685e-05,
      "loss": 1.208,
      "step": 16000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.200874081826463e-05,
      "loss": 1.1678,
      "step": 16500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.176647867124058e-05,
      "loss": 1.2011,
      "step": 17000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.152421652421652e-05,
      "loss": 1.1671,
      "step": 17500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1281954377192475e-05,
      "loss": 1.1849,
      "step": 18000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.103969223016843e-05,
      "loss": 1.1494,
      "step": 18500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0797430083144373e-05,
      "loss": 1.1776,
      "step": 19000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0555652460414365e-05,
      "loss": 1.1664,
      "step": 19500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.031339031339032e-05,
      "loss": 1.1332,
      "step": 20000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.007112816636626e-05,
      "loss": 1.1355,
      "step": 20500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.982886601934221e-05,
      "loss": 1.1624,
      "step": 21000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.958660387231816e-05,
      "loss": 1.1412,
      "step": 21500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.934482624958816e-05,
      "loss": 1.1366,
      "step": 22000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9102564102564105e-05,
      "loss": 1.1227,
      "step": 22500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.886030195554005e-05,
      "loss": 1.0943,
      "step": 23000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.8618039808515996e-05,
      "loss": 1.1645,
      "step": 23500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8376262185786e-05,
      "loss": 1.0841,
      "step": 24000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8134000038761947e-05,
      "loss": 1.118,
      "step": 24500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.789173789173789e-05,
      "loss": 1.1283,
      "step": 25000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.764996026900789e-05,
      "loss": 1.1372,
      "step": 25500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7407698121983836e-05,
      "loss": 1.1181,
      "step": 26000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.716543597495978e-05,
      "loss": 1.0918,
      "step": 26500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6923173827935734e-05,
      "loss": 1.1293,
      "step": 27000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.668139620520573e-05,
      "loss": 1.1014,
      "step": 27500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.643913405818168e-05,
      "loss": 1.0866,
      "step": 28000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6196871911157624e-05,
      "loss": 1.1249,
      "step": 28500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.5954609764133576e-05,
      "loss": 1.1053,
      "step": 29000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5712832141403574e-05,
      "loss": 1.1069,
      "step": 29500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.547056999437952e-05,
      "loss": 1.0803,
      "step": 30000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5228307847355466e-05,
      "loss": 1.1037,
      "step": 30500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.498604570033142e-05,
      "loss": 1.0828,
      "step": 31000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4743783553307364e-05,
      "loss": 1.0805,
      "step": 31500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.450152140628331e-05,
      "loss": 1.0808,
      "step": 32000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.425925925925926e-05,
      "loss": 1.0729,
      "step": 32500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.401699711223521e-05,
      "loss": 1.0854,
      "step": 33000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.3775219489505206e-05,
      "loss": 1.0679,
      "step": 33500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.353295734248115e-05,
      "loss": 1.0779,
      "step": 34000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.32906951954571e-05,
      "loss": 1.0602,
      "step": 34500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.304843304843305e-05,
      "loss": 1.0138,
      "step": 35000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.2806170901409e-05,
      "loss": 1.0125,
      "step": 35500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.256390875438495e-05,
      "loss": 0.9952,
      "step": 36000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2321646607360894e-05,
      "loss": 1.0044,
      "step": 36500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.207986898463089e-05,
      "loss": 0.9936,
      "step": 37000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.183809136190088e-05,
      "loss": 0.9693,
      "step": 37500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.159631373917089e-05,
      "loss": 0.9895,
      "step": 38000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.135405159214683e-05,
      "loss": 0.9724,
      "step": 38500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.111178944512278e-05,
      "loss": 0.9874,
      "step": 39000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.0869527298098725e-05,
      "loss": 0.9774,
      "step": 39500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.062726515107468e-05,
      "loss": 1.0123,
      "step": 40000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0385003004050626e-05,
      "loss": 0.9812,
      "step": 40500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0142740857026575e-05,
      "loss": 0.9907,
      "step": 41000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.990047871000252e-05,
      "loss": 0.9689,
      "step": 41500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9658701087272516e-05,
      "loss": 0.973,
      "step": 42000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9416438940248465e-05,
      "loss": 0.9695,
      "step": 42500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.917417679322441e-05,
      "loss": 0.9766,
      "step": 43000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8931914646200363e-05,
      "loss": 0.9894,
      "step": 43500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8689652499176312e-05,
      "loss": 0.9782,
      "step": 44000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8447390352152258e-05,
      "loss": 0.9702,
      "step": 44500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8205128205128207e-05,
      "loss": 0.9467,
      "step": 45000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.79633505823982e-05,
      "loss": 0.9853,
      "step": 45500
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7721088435374147e-05,
      "loss": 0.9716,
      "step": 46000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.74788262883501e-05,
      "loss": 0.965,
      "step": 46500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.723656414132605e-05,
      "loss": 0.9462,
      "step": 47000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6994301994301995e-05,
      "loss": 0.9536,
      "step": 47500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6752039847277944e-05,
      "loss": 0.9722,
      "step": 48000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.650977770025389e-05,
      "loss": 0.9591,
      "step": 48500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.626751555322984e-05,
      "loss": 0.9715,
      "step": 49000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.602525340620579e-05,
      "loss": 0.9812,
      "step": 49500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5782991259181737e-05,
      "loss": 0.971,
      "step": 50000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5540729112157686e-05,
      "loss": 0.9637,
      "step": 50500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.529895148942768e-05,
      "loss": 0.9404,
      "step": 51000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5056689342403626e-05,
      "loss": 0.9492,
      "step": 51500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.481442719537958e-05,
      "loss": 0.9407,
      "step": 52000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4572165048355525e-05,
      "loss": 0.9618,
      "step": 52500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4329902901331474e-05,
      "loss": 0.9624,
      "step": 53000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4087640754307423e-05,
      "loss": 0.9375,
      "step": 53500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.384537860728337e-05,
      "loss": 0.9463,
      "step": 54000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.360311646025932e-05,
      "loss": 0.9437,
      "step": 54500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3360854313235267e-05,
      "loss": 0.9519,
      "step": 55000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3119561214799313e-05,
      "loss": 0.9356,
      "step": 55500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.287729906777526e-05,
      "loss": 0.965,
      "step": 56000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.263503692075121e-05,
      "loss": 0.947,
      "step": 56500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2392774773727157e-05,
      "loss": 0.9345,
      "step": 57000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2150512626703103e-05,
      "loss": 0.9439,
      "step": 57500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1908250479679052e-05,
      "loss": 0.9375,
      "step": 58000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1665988332655e-05,
      "loss": 0.9437,
      "step": 58500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1424210709924996e-05,
      "loss": 0.9464,
      "step": 59000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1181948562900945e-05,
      "loss": 0.9455,
      "step": 59500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0939686415876894e-05,
      "loss": 0.9208,
      "step": 60000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.069742426885284e-05,
      "loss": 0.9289,
      "step": 60500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.045516212182879e-05,
      "loss": 0.934,
      "step": 61000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0212899974804738e-05,
      "loss": 0.9216,
      "step": 61500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.9970637827780687e-05,
      "loss": 0.9263,
      "step": 62000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9728375680756633e-05,
      "loss": 0.9057,
      "step": 62500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.948659805802663e-05,
      "loss": 0.9022,
      "step": 63000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9244820435296626e-05,
      "loss": 0.9268,
      "step": 63500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9002558288272575e-05,
      "loss": 0.9117,
      "step": 64000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8760296141248524e-05,
      "loss": 0.9253,
      "step": 64500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.851803399422447e-05,
      "loss": 0.937,
      "step": 65000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8275771847200422e-05,
      "loss": 0.9068,
      "step": 65500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8033994224470417e-05,
      "loss": 0.9208,
      "step": 66000
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7791732077446362e-05,
      "loss": 0.9093,
      "step": 66500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7549469930422315e-05,
      "loss": 0.9268,
      "step": 67000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.730720778339826e-05,
      "loss": 0.911,
      "step": 67500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7064945636374206e-05,
      "loss": 0.906,
      "step": 68000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.682268348935016e-05,
      "loss": 0.9129,
      "step": 68500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6580421342326105e-05,
      "loss": 0.8908,
      "step": 69000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.63386437195961e-05,
      "loss": 0.8648,
      "step": 69500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6096866096866097e-05,
      "loss": 0.8501,
      "step": 70000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5854603949842046e-05,
      "loss": 0.8706,
      "step": 70500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5612341802817995e-05,
      "loss": 0.8721,
      "step": 71000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.537007965579394e-05,
      "loss": 0.8385,
      "step": 71500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.512781750876989e-05,
      "loss": 0.8577,
      "step": 72000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.488555536174584e-05,
      "loss": 0.8557,
      "step": 72500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4643293214721787e-05,
      "loss": 0.87,
      "step": 73000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4401031067697734e-05,
      "loss": 0.8454,
      "step": 73500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4158768920673685e-05,
      "loss": 0.8517,
      "step": 74000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3916506773649632e-05,
      "loss": 0.8287,
      "step": 74500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3674729150919627e-05,
      "loss": 0.8464,
      "step": 75000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3432467003895578e-05,
      "loss": 0.8392,
      "step": 75500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3190204856871523e-05,
      "loss": 0.8342,
      "step": 76000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.294794270984747e-05,
      "loss": 0.8241,
      "step": 76500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2706165087117469e-05,
      "loss": 0.8402,
      "step": 77000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2463902940093416e-05,
      "loss": 0.8388,
      "step": 77500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2221640793069365e-05,
      "loss": 0.8416,
      "step": 78000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1979378646045314e-05,
      "loss": 0.8356,
      "step": 78500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1737116499021262e-05,
      "loss": 0.8503,
      "step": 79000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.149485435199721e-05,
      "loss": 0.8415,
      "step": 79500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1252592204973158e-05,
      "loss": 0.8241,
      "step": 80000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1010330057949106e-05,
      "loss": 0.8365,
      "step": 80500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0768552435219102e-05,
      "loss": 0.8324,
      "step": 81000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0526290288195051e-05,
      "loss": 0.8412,
      "step": 81500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0284028141170999e-05,
      "loss": 0.8392,
      "step": 82000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0041765994146948e-05,
      "loss": 0.8169,
      "step": 82500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.799503847122895e-06,
      "loss": 0.8433,
      "step": 83000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.557726224392891e-06,
      "loss": 0.8257,
      "step": 83500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.315464077368839e-06,
      "loss": 0.8252,
      "step": 84000
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.073201930344788e-06,
      "loss": 0.8558,
      "step": 84500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.831424307614784e-06,
      "loss": 0.8249,
      "step": 85000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.589162160590733e-06,
      "loss": 0.8064,
      "step": 85500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.34690001356668e-06,
      "loss": 0.8214,
      "step": 86000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.104637866542628e-06,
      "loss": 0.8255,
      "step": 86500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.862375719518577e-06,
      "loss": 0.8239,
      "step": 87000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.620113572494525e-06,
      "loss": 0.8037,
      "step": 87500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.377851425470474e-06,
      "loss": 0.8353,
      "step": 88000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.135589278446422e-06,
      "loss": 0.8309,
      "step": 88500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.8933271314223696e-06,
      "loss": 0.821,
      "step": 89000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.651064984398319e-06,
      "loss": 0.8332,
      "step": 89500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.408802837374265e-06,
      "loss": 0.8259,
      "step": 90000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.167025214644262e-06,
      "loss": 0.8015,
      "step": 90500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.925247591914259e-06,
      "loss": 0.7962,
      "step": 91000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.682985444890207e-06,
      "loss": 0.8102,
      "step": 91500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.440723297866155e-06,
      "loss": 0.8199,
      "step": 92000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.198461150842103e-06,
      "loss": 0.8308,
      "step": 92500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.956199003818052e-06,
      "loss": 0.8337,
      "step": 93000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.713936856794e-06,
      "loss": 0.8267,
      "step": 93500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.471674709769948e-06,
      "loss": 0.8234,
      "step": 94000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.2294125627458965e-06,
      "loss": 0.8067,
      "step": 94500
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.987150415721844e-06,
      "loss": 0.8037,
      "step": 95000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.7453727929918406e-06,
      "loss": 0.8173,
      "step": 95500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.503110645967789e-06,
      "loss": 0.8107,
      "step": 96000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.2608484989437376e-06,
      "loss": 0.8367,
      "step": 96500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0185863519196854e-06,
      "loss": 0.8082,
      "step": 97000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7763242048956337e-06,
      "loss": 0.8369,
      "step": 97500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.5340620578715816e-06,
      "loss": 0.8116,
      "step": 98000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.2917999108475303e-06,
      "loss": 0.8298,
      "step": 98500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.049537763823478e-06,
      "loss": 0.8025,
      "step": 99000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8072756167994264e-06,
      "loss": 0.8259,
      "step": 99500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5654979940694228e-06,
      "loss": 0.8255,
      "step": 100000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.323235847045371e-06,
      "loss": 0.8258,
      "step": 100500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0814582243153673e-06,
      "loss": 0.8392,
      "step": 101000
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.391960772913154e-07,
      "loss": 0.8188,
      "step": 101500
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.969339302672636e-07,
      "loss": 0.8239,
      "step": 102000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.5467178324321185e-07,
      "loss": 0.8183,
      "step": 102500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1240963621916004e-07,
      "loss": 0.7981,
      "step": 103000
    },
    {
      "epoch": 3.0,
      "step": 103194,
      "total_flos": 1.9167267894224486e+17,
      "train_loss": 0.4842614623898048,
      "train_runtime": 25288.6173,
      "train_samples_per_second": 4.081,
      "train_steps_per_second": 4.081
    }
  ],
  "max_steps": 103194,
  "num_train_epochs": 3,
  "total_flos": 1.9167267894224486e+17,
  "trial_name": null,
  "trial_params": null
}