{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 72552,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.965541956114236e-05,
      "loss": 6.1548,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9310839122284706e-05,
      "loss": 2.7342,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.896625868342706e-05,
      "loss": 2.4512,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8621678244569415e-05,
      "loss": 2.041,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.827709780571177e-05,
      "loss": 1.87,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.793251736685412e-05,
      "loss": 1.788,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7587936927996474e-05,
      "loss": 1.5907,
      "step": 3500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.724335648913883e-05,
      "loss": 1.4284,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.689877605028118e-05,
      "loss": 1.4261,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.655419561142353e-05,
      "loss": 1.3028,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.620961517256589e-05,
      "loss": 1.2614,
      "step": 5500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.586503473370824e-05,
      "loss": 1.22,
      "step": 6000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.552045429485059e-05,
      "loss": 1.1096,
      "step": 6500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5175873855992945e-05,
      "loss": 1.0037,
      "step": 7000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.48312934171353e-05,
      "loss": 0.9759,
      "step": 7500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.448671297827765e-05,
      "loss": 0.9835,
      "step": 8000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.414213253942e-05,
      "loss": 1.0102,
      "step": 8500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.379755210056236e-05,
      "loss": 0.9729,
      "step": 9000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.345297166170471e-05,
      "loss": 0.9993,
      "step": 9500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.310839122284706e-05,
      "loss": 0.9757,
      "step": 10000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2763810783989416e-05,
      "loss": 0.8614,
      "step": 10500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.241923034513177e-05,
      "loss": 0.9695,
      "step": 11000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.207464990627412e-05,
      "loss": 0.9486,
      "step": 11500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1730069467416474e-05,
      "loss": 0.8465,
      "step": 12000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.138548902855883e-05,
      "loss": 0.9047,
      "step": 12500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.1040908589701184e-05,
      "loss": 0.8987,
      "step": 13000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.069632815084353e-05,
      "loss": 0.8616,
      "step": 13500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.035174771198589e-05,
      "loss": 0.8082,
      "step": 14000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.000716727312824e-05,
      "loss": 0.8222,
      "step": 14500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.966258683427059e-05,
      "loss": 0.8866,
      "step": 15000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9318006395412945e-05,
      "loss": 0.8154,
      "step": 15500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.89734259565553e-05,
      "loss": 0.8205,
      "step": 16000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.8628845517697655e-05,
      "loss": 0.7804,
      "step": 16500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8284265078840004e-05,
      "loss": 0.7578,
      "step": 17000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.793968463998236e-05,
      "loss": 0.703,
      "step": 17500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7595104201124713e-05,
      "loss": 0.7462,
      "step": 18000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.725052376226706e-05,
      "loss": 0.6794,
      "step": 18500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.690594332340942e-05,
      "loss": 0.8277,
      "step": 19000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.656136288455177e-05,
      "loss": 0.7999,
      "step": 19500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6216782445694127e-05,
      "loss": 0.8127,
      "step": 20000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5872202006836475e-05,
      "loss": 0.7502,
      "step": 20500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.552762156797883e-05,
      "loss": 0.6666,
      "step": 21000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5183041129121185e-05,
      "loss": 0.7633,
      "step": 21500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.483846069026353e-05,
      "loss": 0.6219,
      "step": 22000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.449388025140589e-05,
      "loss": 0.6019,
      "step": 22500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.414929981254824e-05,
      "loss": 0.6275,
      "step": 23000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.38047193736906e-05,
      "loss": 0.7371,
      "step": 23500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3460138934832946e-05,
      "loss": 0.6864,
      "step": 24000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.31155584959753e-05,
      "loss": 0.5862,
      "step": 24500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.2770978057117656e-05,
      "loss": 0.5982,
      "step": 25000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2426397618260004e-05,
      "loss": 0.6,
      "step": 25500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.2081817179402366e-05,
      "loss": 0.6196,
      "step": 26000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1737236740544714e-05,
      "loss": 0.5537,
      "step": 26500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.139265630168707e-05,
      "loss": 0.5919,
      "step": 27000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.104807586282942e-05,
      "loss": 0.575,
      "step": 27500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.070349542397177e-05,
      "loss": 0.6071,
      "step": 28000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0358914985114127e-05,
      "loss": 0.5724,
      "step": 28500
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.001433454625648e-05,
      "loss": 0.6363,
      "step": 29000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9669754107398834e-05,
      "loss": 0.6006,
      "step": 29500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9325173668541185e-05,
      "loss": 0.5794,
      "step": 30000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.898059322968354e-05,
      "loss": 0.5238,
      "step": 30500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.863601279082589e-05,
      "loss": 0.5694,
      "step": 31000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8291432351968243e-05,
      "loss": 0.5529,
      "step": 31500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.79468519131106e-05,
      "loss": 0.5422,
      "step": 32000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.760227147425295e-05,
      "loss": 0.5753,
      "step": 32500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7257691035395305e-05,
      "loss": 0.5738,
      "step": 33000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6913110596537656e-05,
      "loss": 0.5493,
      "step": 33500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.656853015768001e-05,
      "loss": 0.4863,
      "step": 34000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.622394971882236e-05,
      "loss": 0.5564,
      "step": 34500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5879369279964715e-05,
      "loss": 0.5389,
      "step": 35000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.553478884110707e-05,
      "loss": 0.5948,
      "step": 35500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.519020840224942e-05,
      "loss": 0.5287,
      "step": 36000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4845627963391776e-05,
      "loss": 0.57,
      "step": 36500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4501047524534128e-05,
      "loss": 0.5498,
      "step": 37000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4156467085676483e-05,
      "loss": 0.5333,
      "step": 37500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3811886646818834e-05,
      "loss": 0.5302,
      "step": 38000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3467306207961186e-05,
      "loss": 0.5425,
      "step": 38500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.312272576910354e-05,
      "loss": 0.5765,
      "step": 39000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2778145330245892e-05,
      "loss": 0.5338,
      "step": 39500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2433564891388247e-05,
      "loss": 0.5722,
      "step": 40000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.20889844525306e-05,
      "loss": 0.5721,
      "step": 40500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1744404013672954e-05,
      "loss": 0.5453,
      "step": 41000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1399823574815305e-05,
      "loss": 0.5486,
      "step": 41500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1055243135957657e-05,
      "loss": 0.5039,
      "step": 42000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0710662697100012e-05,
      "loss": 0.5343,
      "step": 42500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0366082258242364e-05,
      "loss": 0.5251,
      "step": 43000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.002150181938472e-05,
      "loss": 0.5104,
      "step": 43500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9676921380527074e-05,
      "loss": 0.5306,
      "step": 44000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9332340941669425e-05,
      "loss": 0.5365,
      "step": 44500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8987760502811777e-05,
      "loss": 0.5759,
      "step": 45000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8643180063954128e-05,
      "loss": 0.5443,
      "step": 45500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8298599625096483e-05,
      "loss": 0.5431,
      "step": 46000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7954019186238835e-05,
      "loss": 0.5675,
      "step": 46500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.760943874738119e-05,
      "loss": 0.5085,
      "step": 47000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7264858308523545e-05,
      "loss": 0.4524,
      "step": 47500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6920277869665896e-05,
      "loss": 0.5196,
      "step": 48000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6575697430808248e-05,
      "loss": 0.4268,
      "step": 48500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6231116991950603e-05,
      "loss": 0.4463,
      "step": 49000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5886536553092954e-05,
      "loss": 0.441,
      "step": 49500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5541956114235306e-05,
      "loss": 0.4984,
      "step": 50000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5197375675377663e-05,
      "loss": 0.4393,
      "step": 50500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4852795236520014e-05,
      "loss": 0.4646,
      "step": 51000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4508214797662368e-05,
      "loss": 0.4159,
      "step": 51500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4163634358804719e-05,
      "loss": 0.4288,
      "step": 52000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3819053919947072e-05,
      "loss": 0.4413,
      "step": 52500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3474473481089426e-05,
      "loss": 0.4242,
      "step": 53000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3129893042231777e-05,
      "loss": 0.4322,
      "step": 53500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2785312603374134e-05,
      "loss": 0.4472,
      "step": 54000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2440732164516485e-05,
      "loss": 0.4213,
      "step": 54500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2096151725658839e-05,
      "loss": 0.4041,
      "step": 55000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1751571286801192e-05,
      "loss": 0.4678,
      "step": 55500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1406990847943544e-05,
      "loss": 0.4285,
      "step": 56000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1062410409085899e-05,
      "loss": 0.4884,
      "step": 56500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.071782997022825e-05,
      "loss": 0.3986,
      "step": 57000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0373249531370603e-05,
      "loss": 0.4442,
      "step": 57500
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0028669092512957e-05,
      "loss": 0.4132,
      "step": 58000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.68408865365531e-06,
      "loss": 0.4151,
      "step": 58500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.339508214797663e-06,
      "loss": 0.4915,
      "step": 59000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.994927775940015e-06,
      "loss": 0.4656,
      "step": 59500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.65034733708237e-06,
      "loss": 0.4357,
      "step": 60000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.305766898224721e-06,
      "loss": 0.4376,
      "step": 60500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.961186459367075e-06,
      "loss": 0.4089,
      "step": 61000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.616606020509429e-06,
      "loss": 0.396,
      "step": 61500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.272025581651781e-06,
      "loss": 0.4529,
      "step": 62000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.927445142794134e-06,
      "loss": 0.3778,
      "step": 62500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.582864703936487e-06,
      "loss": 0.4486,
      "step": 63000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.238284265078841e-06,
      "loss": 0.5516,
      "step": 63500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.8937038262211935e-06,
      "loss": 0.4471,
      "step": 64000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.549123387363547e-06,
      "loss": 0.4037,
      "step": 64500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.204542948505899e-06,
      "loss": 0.4551,
      "step": 65000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.8599625096482524e-06,
      "loss": 0.3944,
      "step": 65500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.515382070790605e-06,
      "loss": 0.4431,
      "step": 66000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.170801631932959e-06,
      "loss": 0.4195,
      "step": 66500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.826221193075312e-06,
      "loss": 0.4728,
      "step": 67000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.4816407542176647e-06,
      "loss": 0.4298,
      "step": 67500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.137060315360018e-06,
      "loss": 0.4228,
      "step": 68000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.792479876502371e-06,
      "loss": 0.4302,
      "step": 68500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.447899437644724e-06,
      "loss": 0.4204,
      "step": 69000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.103318998787077e-06,
      "loss": 0.4077,
      "step": 69500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.75873855992943e-06,
      "loss": 0.4172,
      "step": 70000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.414158121071783e-06,
      "loss": 0.389,
      "step": 70500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.069577682214136e-06,
      "loss": 0.4612,
      "step": 71000
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.249972433564892e-07,
      "loss": 0.447,
      "step": 71500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.8041680449884223e-07,
      "loss": 0.4153,
      "step": 72000
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.5836365641195283e-08,
      "loss": 0.4197,
      "step": 72500
    },
    {
      "epoch": 3.0,
      "step": 72552,
      "total_flos": 6412401164344320.0,
      "train_loss": 0.7122196425333716,
      "train_runtime": 31286.0654,
      "train_samples_per_second": 2.319,
      "train_steps_per_second": 2.319
    }
  ],
  "max_steps": 72552,
  "num_train_epochs": 3,
  "total_flos": 6412401164344320.0,
  "trial_name": null,
  "trial_params": null
}