{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 32740,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15,
      "learning_rate": 9.847281612706171e-06,
      "loss": 230560.032,
      "step": 500
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.69456322541234e-06,
      "loss": 0.4568,
      "step": 1000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.541844838118511e-06,
      "loss": 0.4604,
      "step": 1500
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.389126450824681e-06,
      "loss": 0.4138,
      "step": 2000
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.23640806353085e-06,
      "loss": 0.3819,
      "step": 2500
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.08368967623702e-06,
      "loss": 0.3615,
      "step": 3000
    },
    {
      "epoch": 1.07,
      "learning_rate": 8.93097128894319e-06,
      "loss": 0.348,
      "step": 3500
    },
    {
      "epoch": 1.22,
      "learning_rate": 8.77825290164936e-06,
      "loss": 0.4017,
      "step": 4000
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.62553451435553e-06,
      "loss": 0.3136,
      "step": 4500
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.4728161270617e-06,
      "loss": 0.3419,
      "step": 5000
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.320097739767868e-06,
      "loss": 0.3969,
      "step": 5500
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.167379352474038e-06,
      "loss": 0.304,
      "step": 6000
    },
    {
      "epoch": 1.99,
      "learning_rate": 8.014660965180208e-06,
      "loss": 0.2897,
      "step": 6500
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.861942577886378e-06,
      "loss": 0.2875,
      "step": 7000
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.709224190592548e-06,
      "loss": 0.2658,
      "step": 7500
    },
    {
      "epoch": 2.44,
      "learning_rate": 7.556505803298718e-06,
      "loss": 0.2658,
      "step": 8000
    },
    {
      "epoch": 2.6,
      "learning_rate": 7.403787416004888e-06,
      "loss": 0.2861,
      "step": 8500
    },
    {
      "epoch": 2.75,
      "learning_rate": 7.251069028711058e-06,
      "loss": 0.2862,
      "step": 9000
    },
    {
      "epoch": 2.9,
      "learning_rate": 7.098350641417228e-06,
      "loss": 0.2594,
      "step": 9500
    },
    {
      "epoch": 3.05,
      "learning_rate": 6.945632254123397e-06,
      "loss": 0.2509,
      "step": 10000
    },
    {
      "epoch": 3.21,
      "learning_rate": 6.792913866829567e-06,
      "loss": 0.2535,
      "step": 10500
    },
    {
      "epoch": 3.36,
      "learning_rate": 6.640195479535737e-06,
      "loss": 0.2358,
      "step": 11000
    },
    {
      "epoch": 3.51,
      "learning_rate": 6.487477092241906e-06,
      "loss": 0.2357,
      "step": 11500
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.334758704948076e-06,
      "loss": 0.2331,
      "step": 12000
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.1820403176542464e-06,
      "loss": 0.2806,
      "step": 12500
    },
    {
      "epoch": 3.97,
      "learning_rate": 6.029321930360416e-06,
      "loss": 0.2245,
      "step": 13000
    },
    {
      "epoch": 4.12,
      "learning_rate": 5.876603543066586e-06,
      "loss": 0.2074,
      "step": 13500
    },
    {
      "epoch": 4.28,
      "learning_rate": 5.723885155772756e-06,
      "loss": 0.2115,
      "step": 14000
    },
    {
      "epoch": 4.43,
      "learning_rate": 5.571166768478925e-06,
      "loss": 0.2124,
      "step": 14500
    },
    {
      "epoch": 4.58,
      "learning_rate": 5.418448381185095e-06,
      "loss": 0.2149,
      "step": 15000
    },
    {
      "epoch": 4.73,
      "learning_rate": 5.265729993891265e-06,
      "loss": 0.2133,
      "step": 15500
    },
    {
      "epoch": 4.89,
      "learning_rate": 5.113011606597434e-06,
      "loss": 0.2119,
      "step": 16000
    },
    {
      "epoch": 5.04,
      "learning_rate": 4.960293219303604e-06,
      "loss": 0.2128,
      "step": 16500
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.8075748320097744e-06,
      "loss": 0.1881,
      "step": 17000
    },
    {
      "epoch": 5.35,
      "learning_rate": 4.6548564447159445e-06,
      "loss": 0.1974,
      "step": 17500
    },
    {
      "epoch": 5.5,
      "learning_rate": 4.502138057422114e-06,
      "loss": 0.1985,
      "step": 18000
    },
    {
      "epoch": 5.65,
      "learning_rate": 4.349419670128284e-06,
      "loss": 0.1979,
      "step": 18500
    },
    {
      "epoch": 5.8,
      "learning_rate": 4.196701282834454e-06,
      "loss": 0.1986,
      "step": 19000
    },
    {
      "epoch": 5.96,
      "learning_rate": 4.043982895540624e-06,
      "loss": 0.1906,
      "step": 19500
    },
    {
      "epoch": 6.11,
      "learning_rate": 3.891264508246793e-06,
      "loss": 0.1873,
      "step": 20000
    },
    {
      "epoch": 6.26,
      "learning_rate": 3.7385461209529628e-06,
      "loss": 0.1832,
      "step": 20500
    },
    {
      "epoch": 6.41,
      "learning_rate": 3.585827733659133e-06,
      "loss": 0.1786,
      "step": 21000
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.4331093463653025e-06,
      "loss": 0.1811,
      "step": 21500
    },
    {
      "epoch": 6.72,
      "learning_rate": 3.280390959071472e-06,
      "loss": 0.1856,
      "step": 22000
    },
    {
      "epoch": 6.87,
      "learning_rate": 3.1276725717776426e-06,
      "loss": 0.1767,
      "step": 22500
    },
    {
      "epoch": 7.03,
      "learning_rate": 2.974954184483812e-06,
      "loss": 0.1744,
      "step": 23000
    },
    {
      "epoch": 7.18,
      "learning_rate": 2.8222357971899814e-06,
      "loss": 0.17,
      "step": 23500
    },
    {
      "epoch": 7.33,
      "learning_rate": 2.669517409896152e-06,
      "loss": 0.178,
      "step": 24000
    },
    {
      "epoch": 7.48,
      "learning_rate": 2.5167990226023216e-06,
      "loss": 0.1742,
      "step": 24500
    },
    {
      "epoch": 7.64,
      "learning_rate": 2.364080635308491e-06,
      "loss": 0.1673,
      "step": 25000
    },
    {
      "epoch": 7.79,
      "learning_rate": 2.2113622480146613e-06,
      "loss": 0.1634,
      "step": 25500
    },
    {
      "epoch": 7.94,
      "learning_rate": 2.058643860720831e-06,
      "loss": 0.1735,
      "step": 26000
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.9059254734270008e-06,
      "loss": 0.1691,
      "step": 26500
    },
    {
      "epoch": 8.25,
      "learning_rate": 1.7532070861331706e-06,
      "loss": 0.1697,
      "step": 27000
    },
    {
      "epoch": 8.4,
      "learning_rate": 1.6004886988393402e-06,
      "loss": 0.1585,
      "step": 27500
    },
    {
      "epoch": 8.55,
      "learning_rate": 1.4477703115455103e-06,
      "loss": 0.1591,
      "step": 28000
    },
    {
      "epoch": 8.7,
      "learning_rate": 1.2950519242516802e-06,
      "loss": 0.1622,
      "step": 28500
    },
    {
      "epoch": 8.86,
      "learning_rate": 1.1423335369578498e-06,
      "loss": 0.1558,
      "step": 29000
    },
    {
      "epoch": 9.01,
      "learning_rate": 9.896151496640196e-07,
      "loss": 0.1598,
      "step": 29500
    },
    {
      "epoch": 9.16,
      "learning_rate": 8.368967623701894e-07,
      "loss": 0.1515,
      "step": 30000
    },
    {
      "epoch": 9.32,
      "learning_rate": 6.841783750763592e-07,
      "loss": 0.1556,
      "step": 30500
    },
    {
      "epoch": 9.47,
      "learning_rate": 5.314599877825291e-07,
      "loss": 0.1526,
      "step": 31000
    },
    {
      "epoch": 9.62,
      "learning_rate": 3.787416004886989e-07,
      "loss": 0.1586,
      "step": 31500
    },
    {
      "epoch": 9.77,
      "learning_rate": 2.2602321319486868e-07,
      "loss": 0.1489,
      "step": 32000
    },
    {
      "epoch": 9.93,
      "learning_rate": 7.330482590103849e-08,
      "loss": 0.1567,
      "step": 32500
    },
    {
      "epoch": 10.0,
      "step": 32740,
      "total_flos": 1.64206965272832e+16,
      "train_loss": 3521.3033845717396,
      "train_runtime": 13419.3371,
      "train_samples_per_second": 78.054,
      "train_steps_per_second": 2.44
    }
  ],
  "max_steps": 32740,
  "num_train_epochs": 10,
  "total_flos": 1.64206965272832e+16,
  "trial_name": null,
  "trial_params": null
}