{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 135420,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9815389159651454e-05,
      "loss": 2.2712,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.963077831930291e-05,
      "loss": 1.9301,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9446167478954364e-05,
      "loss": 1.7662,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.926155663860582e-05,
      "loss": 1.6799,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9076945798257274e-05,
      "loss": 1.5865,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.889233495790873e-05,
      "loss": 1.5379,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8707724117560184e-05,
      "loss": 1.4763,
      "step": 3500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.852311327721164e-05,
      "loss": 1.4542,
      "step": 4000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8338502436863094e-05,
      "loss": 1.4167,
      "step": 4500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.815389159651455e-05,
      "loss": 1.385,
      "step": 5000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7969280756166004e-05,
      "loss": 1.3705,
      "step": 5500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.778466991581746e-05,
      "loss": 1.343,
      "step": 6000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7600059075468914e-05,
      "loss": 1.328,
      "step": 6500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7415448235120366e-05,
      "loss": 1.3042,
      "step": 7000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7230837394771824e-05,
      "loss": 1.282,
      "step": 7500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7046226554423276e-05,
      "loss": 1.2674,
      "step": 8000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.686161571407473e-05,
      "loss": 1.266,
      "step": 8500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6677004873726186e-05,
      "loss": 1.2439,
      "step": 9000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.649239403337764e-05,
      "loss": 1.2345,
      "step": 9500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6307783193029096e-05,
      "loss": 1.2208,
      "step": 10000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.612317235268055e-05,
      "loss": 1.2138,
      "step": 10500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.5938561512332006e-05,
      "loss": 1.2013,
      "step": 11000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.575395067198346e-05,
      "loss": 1.1848,
      "step": 11500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5569339831634916e-05,
      "loss": 1.1898,
      "step": 12000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.538472899128637e-05,
      "loss": 1.1613,
      "step": 12500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5200118150937826e-05,
      "loss": 1.1654,
      "step": 13000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.501550731058928e-05,
      "loss": 1.159,
      "step": 13500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4830896470240736e-05,
      "loss": 1.1474,
      "step": 14000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.464628562989219e-05,
      "loss": 1.1353,
      "step": 14500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4461674789543646e-05,
      "loss": 1.1443,
      "step": 15000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.42770639491951e-05,
      "loss": 1.1293,
      "step": 15500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4092453108846556e-05,
      "loss": 1.1075,
      "step": 16000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.390784226849801e-05,
      "loss": 1.112,
      "step": 16500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.3723231428149466e-05,
      "loss": 1.1072,
      "step": 17000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.353862058780092e-05,
      "loss": 1.1057,
      "step": 17500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3354009747452376e-05,
      "loss": 1.0928,
      "step": 18000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.316939890710383e-05,
      "loss": 1.0815,
      "step": 18500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.2984788066755286e-05,
      "loss": 1.1019,
      "step": 19000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.280017722640674e-05,
      "loss": 1.0808,
      "step": 19500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.261556638605819e-05,
      "loss": 1.0799,
      "step": 20000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.243095554570964e-05,
      "loss": 1.0777,
      "step": 20500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.22463447053611e-05,
      "loss": 1.0864,
      "step": 21000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.206173386501255e-05,
      "loss": 1.05,
      "step": 21500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.187712302466401e-05,
      "loss": 1.0586,
      "step": 22000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.169251218431546e-05,
      "loss": 1.0696,
      "step": 22500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.150790134396692e-05,
      "loss": 1.0415,
      "step": 23000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.132329050361837e-05,
      "loss": 1.0549,
      "step": 23500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.113867966326983e-05,
      "loss": 1.0401,
      "step": 24000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.095406882292128e-05,
      "loss": 1.0386,
      "step": 24500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.076945798257274e-05,
      "loss": 1.0303,
      "step": 25000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.058484714222419e-05,
      "loss": 1.0258,
      "step": 25500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.040023630187565e-05,
      "loss": 1.0435,
      "step": 26000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.02156254615271e-05,
      "loss": 1.0201,
      "step": 26500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.003101462117856e-05,
      "loss": 1.0257,
      "step": 27000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.984640378083001e-05,
      "loss": 1.0205,
      "step": 27500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.966179294048147e-05,
      "loss": 1.025,
      "step": 28000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.947718210013292e-05,
      "loss": 1.0109,
      "step": 28500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.929257125978438e-05,
      "loss": 1.0164,
      "step": 29000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.910796041943583e-05,
      "loss": 1.0185,
      "step": 29500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.892334957908729e-05,
      "loss": 1.0017,
      "step": 30000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.873873873873874e-05,
      "loss": 0.9901,
      "step": 30500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.85541278983902e-05,
      "loss": 1.0056,
      "step": 31000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.836951705804165e-05,
      "loss": 0.9878,
      "step": 31500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.818490621769311e-05,
      "loss": 0.9905,
      "step": 32000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.800029537734456e-05,
      "loss": 0.9977,
      "step": 32500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.781568453699601e-05,
      "loss": 0.985,
      "step": 33000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7631073696647464e-05,
      "loss": 0.9838,
      "step": 33500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.744646285629892e-05,
      "loss": 0.9837,
      "step": 34000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7261852015950374e-05,
      "loss": 0.9857,
      "step": 34500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.707724117560183e-05,
      "loss": 0.979,
      "step": 35000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6892630335253284e-05,
      "loss": 0.9859,
      "step": 35500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.670801949490474e-05,
      "loss": 0.9666,
      "step": 36000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6523408654556194e-05,
      "loss": 0.9747,
      "step": 36500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.633879781420765e-05,
      "loss": 0.9689,
      "step": 37000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6154186973859104e-05,
      "loss": 0.9679,
      "step": 37500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.596957613351056e-05,
      "loss": 0.9576,
      "step": 38000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5784965293162014e-05,
      "loss": 0.9568,
      "step": 38500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.560035445281347e-05,
      "loss": 0.9661,
      "step": 39000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5415743612464924e-05,
      "loss": 0.9587,
      "step": 39500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.523113277211638e-05,
      "loss": 0.9417,
      "step": 40000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.5046521931767834e-05,
      "loss": 0.954,
      "step": 40500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.486191109141929e-05,
      "loss": 0.952,
      "step": 41000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4677300251070744e-05,
      "loss": 0.9526,
      "step": 41500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.44926894107222e-05,
      "loss": 0.9406,
      "step": 42000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4308078570373654e-05,
      "loss": 0.9526,
      "step": 42500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.412346773002511e-05,
      "loss": 0.9475,
      "step": 43000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3938856889676564e-05,
      "loss": 0.945,
      "step": 43500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.375424604932802e-05,
      "loss": 0.9415,
      "step": 44000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3569635208979474e-05,
      "loss": 0.94,
      "step": 44500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3385024368630926e-05,
      "loss": 0.9362,
      "step": 45000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3200413528282384e-05,
      "loss": 0.8705,
      "step": 45500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3015802687933836e-05,
      "loss": 0.8512,
      "step": 46000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.283119184758529e-05,
      "loss": 0.8513,
      "step": 46500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.2646581007236746e-05,
      "loss": 0.856,
      "step": 47000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.24619701668882e-05,
      "loss": 0.8546,
      "step": 47500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2277359326539656e-05,
      "loss": 0.8532,
      "step": 48000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.209274848619111e-05,
      "loss": 0.8487,
      "step": 48500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.1908137645842566e-05,
      "loss": 0.8411,
      "step": 49000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.172352680549402e-05,
      "loss": 0.8444,
      "step": 49500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.1538915965145476e-05,
      "loss": 0.8485,
      "step": 50000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.135430512479693e-05,
      "loss": 0.8469,
      "step": 50500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1169694284448386e-05,
      "loss": 0.8435,
      "step": 51000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.098508344409984e-05,
      "loss": 0.8556,
      "step": 51500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.0800472603751296e-05,
      "loss": 0.845,
      "step": 52000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.061586176340275e-05,
      "loss": 0.8437,
      "step": 52500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0431250923054206e-05,
      "loss": 0.8403,
      "step": 53000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0246640082705658e-05,
      "loss": 0.8336,
      "step": 53500
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.0062029242357116e-05,
      "loss": 0.8396,
      "step": 54000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9877418402008568e-05,
      "loss": 0.8408,
      "step": 54500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9692807561660023e-05,
      "loss": 0.8429,
      "step": 55000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9508196721311478e-05,
      "loss": 0.8366,
      "step": 55500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9323585880962933e-05,
      "loss": 0.8414,
      "step": 56000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9138975040614384e-05,
      "loss": 0.8464,
      "step": 56500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8954364200265843e-05,
      "loss": 0.8164,
      "step": 57000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8769753359917294e-05,
      "loss": 0.8353,
      "step": 57500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8585142519568753e-05,
      "loss": 0.8431,
      "step": 58000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8400531679220204e-05,
      "loss": 0.8407,
      "step": 58500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8215920838871663e-05,
      "loss": 0.8313,
      "step": 59000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8031309998523115e-05,
      "loss": 0.8323,
      "step": 59500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7846699158174573e-05,
      "loss": 0.8284,
      "step": 60000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7662088317826025e-05,
      "loss": 0.8372,
      "step": 60500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7477477477477483e-05,
      "loss": 0.8274,
      "step": 61000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7292866637128935e-05,
      "loss": 0.831,
      "step": 61500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.710825579678039e-05,
      "loss": 0.8221,
      "step": 62000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.692364495643184e-05,
      "loss": 0.8168,
      "step": 62500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.67390341160833e-05,
      "loss": 0.8172,
      "step": 63000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.655442327573475e-05,
      "loss": 0.8189,
      "step": 63500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.636981243538621e-05,
      "loss": 0.8275,
      "step": 64000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.618520159503766e-05,
      "loss": 0.829,
      "step": 64500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.600059075468912e-05,
      "loss": 0.8151,
      "step": 65000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.581597991434057e-05,
      "loss": 0.8126,
      "step": 65500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.563136907399203e-05,
      "loss": 0.824,
      "step": 66000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.544675823364348e-05,
      "loss": 0.8213,
      "step": 66500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.526214739329494e-05,
      "loss": 0.8153,
      "step": 67000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.507753655294639e-05,
      "loss": 0.8303,
      "step": 67500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4892925712597846e-05,
      "loss": 0.8209,
      "step": 68000
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.47083148722493e-05,
      "loss": 0.835,
      "step": 68500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4523704031900753e-05,
      "loss": 0.8138,
      "step": 69000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4339093191552208e-05,
      "loss": 0.8069,
      "step": 69500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4154482351203663e-05,
      "loss": 0.8132,
      "step": 70000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3969871510855118e-05,
      "loss": 0.814,
      "step": 70500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3785260670506573e-05,
      "loss": 0.8082,
      "step": 71000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3600649830158028e-05,
      "loss": 0.8043,
      "step": 71500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3416038989809483e-05,
      "loss": 0.8095,
      "step": 72000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3231428149460938e-05,
      "loss": 0.8064,
      "step": 72500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3046817309112393e-05,
      "loss": 0.8086,
      "step": 73000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2862206468763848e-05,
      "loss": 0.8096,
      "step": 73500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2677595628415303e-05,
      "loss": 0.808,
      "step": 74000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2492984788066758e-05,
      "loss": 0.8151,
      "step": 74500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2308373947718213e-05,
      "loss": 0.8036,
      "step": 75000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2123763107369665e-05,
      "loss": 0.801,
      "step": 75500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.193915226702112e-05,
      "loss": 0.8111,
      "step": 76000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1754541426672575e-05,
      "loss": 0.8121,
      "step": 76500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.156993058632403e-05,
      "loss": 0.8094,
      "step": 77000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1385319745975485e-05,
      "loss": 0.8017,
      "step": 77500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.120070890562694e-05,
      "loss": 0.7976,
      "step": 78000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1016098065278395e-05,
      "loss": 0.7893,
      "step": 78500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.083148722492985e-05,
      "loss": 0.807,
      "step": 79000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0646876384581305e-05,
      "loss": 0.7964,
      "step": 79500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.046226554423276e-05,
      "loss": 0.7919,
      "step": 80000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0277654703884215e-05,
      "loss": 0.7943,
      "step": 80500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.009304386353567e-05,
      "loss": 0.7927,
      "step": 81000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.990843302318712e-05,
      "loss": 0.7977,
      "step": 81500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9723822182838576e-05,
      "loss": 0.7964,
      "step": 82000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.953921134249003e-05,
      "loss": 0.7953,
      "step": 82500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9354600502141486e-05,
      "loss": 0.8002,
      "step": 83000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.916998966179294e-05,
      "loss": 0.7896,
      "step": 83500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8985378821444396e-05,
      "loss": 0.783,
      "step": 84000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.880076798109585e-05,
      "loss": 0.7865,
      "step": 84500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8616157140747306e-05,
      "loss": 0.7852,
      "step": 85000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.843154630039876e-05,
      "loss": 0.7893,
      "step": 85500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8246935460050216e-05,
      "loss": 0.7919,
      "step": 86000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.806232461970167e-05,
      "loss": 0.7801,
      "step": 86500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7877713779353126e-05,
      "loss": 0.7733,
      "step": 87000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.769310293900458e-05,
      "loss": 0.7764,
      "step": 87500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7508492098656033e-05,
      "loss": 0.7849,
      "step": 88000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7323881258307488e-05,
      "loss": 0.7832,
      "step": 88500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7139270417958943e-05,
      "loss": 0.7788,
      "step": 89000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6954659577610398e-05,
      "loss": 0.7716,
      "step": 89500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6770048737261853e-05,
      "loss": 0.7686,
      "step": 90000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6585437896913308e-05,
      "loss": 0.7509,
      "step": 90500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6400827056564763e-05,
      "loss": 0.7014,
      "step": 91000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6216216216216218e-05,
      "loss": 0.6893,
      "step": 91500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6031605375867673e-05,
      "loss": 0.691,
      "step": 92000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5846994535519128e-05,
      "loss": 0.6965,
      "step": 92500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5662383695170583e-05,
      "loss": 0.6964,
      "step": 93000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5477772854822038e-05,
      "loss": 0.7002,
      "step": 93500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5293162014473493e-05,
      "loss": 0.6945,
      "step": 94000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5108551174124946e-05,
      "loss": 0.6968,
      "step": 94500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4923940333776401e-05,
      "loss": 0.7044,
      "step": 95000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4739329493427856e-05,
      "loss": 0.69,
      "step": 95500
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.455471865307931e-05,
      "loss": 0.6976,
      "step": 96000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4370107812730765e-05,
      "loss": 0.6885,
      "step": 96500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.418549697238222e-05,
      "loss": 0.6895,
      "step": 97000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4000886132033675e-05,
      "loss": 0.693,
      "step": 97500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.381627529168513e-05,
      "loss": 0.6906,
      "step": 98000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3631664451336585e-05,
      "loss": 0.6963,
      "step": 98500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.344705361098804e-05,
      "loss": 0.7016,
      "step": 99000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3262442770639493e-05,
      "loss": 0.6896,
      "step": 99500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3077831930290948e-05,
      "loss": 0.6881,
      "step": 100000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2893221089942403e-05,
      "loss": 0.6908,
      "step": 100500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2708610249593858e-05,
      "loss": 0.6893,
      "step": 101000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2523999409245313e-05,
      "loss": 0.6908,
      "step": 101500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2339388568896766e-05,
      "loss": 0.6966,
      "step": 102000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2154777728548221e-05,
      "loss": 0.6926,
      "step": 102500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1970166888199675e-05,
      "loss": 0.6981,
      "step": 103000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.178555604785113e-05,
      "loss": 0.6911,
      "step": 103500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1600945207502585e-05,
      "loss": 0.6917,
      "step": 104000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.141633436715404e-05,
      "loss": 0.679,
      "step": 104500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1231723526805495e-05,
      "loss": 0.6785,
      "step": 105000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.104711268645695e-05,
      "loss": 0.6804,
      "step": 105500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0862501846108405e-05,
      "loss": 0.6921,
      "step": 106000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0677891005759858e-05,
      "loss": 0.6983,
      "step": 106500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0493280165411313e-05,
      "loss": 0.6831,
      "step": 107000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0308669325062768e-05,
      "loss": 0.6865,
      "step": 107500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0124058484714223e-05,
      "loss": 0.6891,
      "step": 108000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.939447644365678e-06,
      "loss": 0.6858,
      "step": 108500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.754836804017133e-06,
      "loss": 0.683,
      "step": 109000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.570225963668587e-06,
      "loss": 0.6737,
      "step": 109500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.385615123320042e-06,
      "loss": 0.6863,
      "step": 110000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.201004282971497e-06,
      "loss": 0.6884,
      "step": 110500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.016393442622952e-06,
      "loss": 0.6897,
      "step": 111000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.831782602274407e-06,
      "loss": 0.6748,
      "step": 111500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.647171761925862e-06,
      "loss": 0.6813,
      "step": 112000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.462560921577317e-06,
      "loss": 0.683,
      "step": 112500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.27795008122877e-06,
      "loss": 0.6775,
      "step": 113000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.093339240880225e-06,
      "loss": 0.6831,
      "step": 113500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.90872840053168e-06,
      "loss": 0.6904,
      "step": 114000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.724117560183135e-06,
      "loss": 0.6688,
      "step": 114500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.539506719834589e-06,
      "loss": 0.6821,
      "step": 115000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.354895879486044e-06,
      "loss": 0.6711,
      "step": 115500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.170285039137499e-06,
      "loss": 0.6749,
      "step": 116000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.985674198788954e-06,
      "loss": 0.6789,
      "step": 116500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.801063358440408e-06,
      "loss": 0.6722,
      "step": 117000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.616452518091863e-06,
      "loss": 0.6754,
      "step": 117500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.431841677743318e-06,
      "loss": 0.6672,
      "step": 118000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.2472308373947724e-06,
      "loss": 0.6736,
      "step": 118500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.062619997046227e-06,
      "loss": 0.6715,
      "step": 119000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.878009156697682e-06,
      "loss": 0.6677,
      "step": 119500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.693398316349137e-06,
      "loss": 0.6799,
      "step": 120000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.508787476000591e-06,
      "loss": 0.6709,
      "step": 120500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.324176635652046e-06,
      "loss": 0.669,
      "step": 121000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.139565795303501e-06,
      "loss": 0.6712,
      "step": 121500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.954954954954955e-06,
      "loss": 0.6775,
      "step": 122000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.77034411460641e-06,
      "loss": 0.6842,
      "step": 122500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.585733274257865e-06,
      "loss": 0.6744,
      "step": 123000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.401122433909319e-06,
      "loss": 0.6769,
      "step": 123500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.216511593560774e-06,
      "loss": 0.6746,
      "step": 124000
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.031900753212229e-06,
      "loss": 0.6743,
      "step": 124500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.847289912863683e-06,
      "loss": 0.6721,
      "step": 125000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.6626790725151383e-06,
      "loss": 0.6792,
      "step": 125500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.4780682321665933e-06,
      "loss": 0.6754,
      "step": 126000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.293457391818048e-06,
      "loss": 0.6692,
      "step": 126500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1088465514695025e-06,
      "loss": 0.6695,
      "step": 127000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.924235711120957e-06,
      "loss": 0.6652,
      "step": 127500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.7396248707724117e-06,
      "loss": 0.6661,
      "step": 128000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.5550140304238667e-06,
      "loss": 0.6722,
      "step": 128500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3704031900753213e-06,
      "loss": 0.6661,
      "step": 129000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.185792349726776e-06,
      "loss": 0.6714,
      "step": 129500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.001181509378231e-06,
      "loss": 0.6686,
      "step": 130000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8165706690296854e-06,
      "loss": 0.6675,
      "step": 130500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6319598286811402e-06,
      "loss": 0.6666,
      "step": 131000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.447348988332595e-06,
      "loss": 0.6691,
      "step": 131500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2627381479840496e-06,
      "loss": 0.6699,
      "step": 132000
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0781273076355044e-06,
      "loss": 0.663,
      "step": 132500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.935164672869591e-07,
      "loss": 0.6723,
      "step": 133000
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.089056269384138e-07,
      "loss": 0.6687,
      "step": 133500
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.242947865898686e-07,
      "loss": 0.6617,
      "step": 134000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.396839462413233e-07,
      "loss": 0.6632,
      "step": 134500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.5507310589277803e-07,
      "loss": 0.6641,
      "step": 135000
    },
    {
      "epoch": 3.0,
      "step": 135420,
      "total_flos": 3.592406571113513e+17,
      "train_loss": 0.8762090477628628,
      "train_runtime": 47941.5748,
      "train_samples_per_second": 56.493,
      "train_steps_per_second": 2.825
    }
  ],
  "max_steps": 135420,
  "num_train_epochs": 3,
  "total_flos": 3.592406571113513e+17,
  "trial_name": null,
  "trial_params": null
}