{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 36816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 9.864189482833552e-06,
      "loss": 0.9064,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.728378965667103e-06,
      "loss": 0.6939,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.592568448500653e-06,
      "loss": 0.6652,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.456757931334203e-06,
      "loss": 0.6331,
      "step": 2000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.320947414167753e-06,
      "loss": 0.6061,
      "step": 2500
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.185136897001304e-06,
      "loss": 0.5958,
      "step": 3000
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.049326379834855e-06,
      "loss": 0.588,
      "step": 3500
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.913515862668406e-06,
      "loss": 0.5726,
      "step": 4000
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.777705345501957e-06,
      "loss": 0.5618,
      "step": 4500
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.641894828335508e-06,
      "loss": 0.5599,
      "step": 5000
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.506084311169057e-06,
      "loss": 0.5557,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.370273794002608e-06,
      "loss": 0.5444,
      "step": 6000
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.234463276836159e-06,
      "loss": 0.5423,
      "step": 6500
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.09865275966971e-06,
      "loss": 0.5322,
      "step": 7000
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.96284224250326e-06,
      "loss": 0.541,
      "step": 7500
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.827031725336811e-06,
      "loss": 0.5284,
      "step": 8000
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.691221208170362e-06,
      "loss": 0.524,
      "step": 8500
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.555410691003911e-06,
      "loss": 0.5213,
      "step": 9000
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.419600173837462e-06,
      "loss": 0.5153,
      "step": 9500
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.283789656671013e-06,
      "loss": 0.5256,
      "step": 10000
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.147979139504564e-06,
      "loss": 0.509,
      "step": 10500
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.012168622338115e-06,
      "loss": 0.5088,
      "step": 11000
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.8763581051716655e-06,
      "loss": 0.5053,
      "step": 11500
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.7405475880052155e-06,
      "loss": 0.4892,
      "step": 12000
    },
    {
      "epoch": 1.02,
      "learning_rate": 6.6047370708387655e-06,
      "loss": 0.484,
      "step": 12500
    },
    {
      "epoch": 1.06,
      "learning_rate": 6.468926553672316e-06,
      "loss": 0.4504,
      "step": 13000
    },
    {
      "epoch": 1.1,
      "learning_rate": 6.333116036505867e-06,
      "loss": 0.4658,
      "step": 13500
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.197305519339418e-06,
      "loss": 0.4468,
      "step": 14000
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.061495002172969e-06,
      "loss": 0.4542,
      "step": 14500
    },
    {
      "epoch": 1.22,
      "learning_rate": 5.92568448500652e-06,
      "loss": 0.4555,
      "step": 15000
    },
    {
      "epoch": 1.26,
      "learning_rate": 5.78987396784007e-06,
      "loss": 0.4602,
      "step": 15500
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.65406345067362e-06,
      "loss": 0.4465,
      "step": 16000
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.518252933507171e-06,
      "loss": 0.4601,
      "step": 16500
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.382442416340722e-06,
      "loss": 0.4595,
      "step": 17000
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.2466318991742724e-06,
      "loss": 0.4534,
      "step": 17500
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.110821382007823e-06,
      "loss": 0.4464,
      "step": 18000
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.975010864841373e-06,
      "loss": 0.4398,
      "step": 18500
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.839200347674924e-06,
      "loss": 0.4433,
      "step": 19000
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.703389830508475e-06,
      "loss": 0.441,
      "step": 19500
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.567579313342025e-06,
      "loss": 0.4479,
      "step": 20000
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.431768796175576e-06,
      "loss": 0.4471,
      "step": 20500
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.295958279009127e-06,
      "loss": 0.4431,
      "step": 21000
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.160147761842678e-06,
      "loss": 0.4389,
      "step": 21500
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.024337244676228e-06,
      "loss": 0.4404,
      "step": 22000
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.8885267275097785e-06,
      "loss": 0.4431,
      "step": 22500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.7527162103433294e-06,
      "loss": 0.4343,
      "step": 23000
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.61690569317688e-06,
      "loss": 0.4335,
      "step": 23500
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.4810951760104307e-06,
      "loss": 0.4354,
      "step": 24000
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.3452846588439815e-06,
      "loss": 0.4493,
      "step": 24500
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.2094741416775315e-06,
      "loss": 0.3965,
      "step": 25000
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.0736636245110824e-06,
      "loss": 0.4061,
      "step": 25500
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.9378531073446333e-06,
      "loss": 0.4041,
      "step": 26000
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.8020425901781833e-06,
      "loss": 0.3955,
      "step": 26500
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.666232073011734e-06,
      "loss": 0.4015,
      "step": 27000
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.530421555845285e-06,
      "loss": 0.3917,
      "step": 27500
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.3946110386788354e-06,
      "loss": 0.397,
      "step": 28000
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.2588005215123863e-06,
      "loss": 0.3925,
      "step": 28500
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.1229900043459367e-06,
      "loss": 0.3933,
      "step": 29000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.987179487179487e-06,
      "loss": 0.404,
      "step": 29500
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.851368970013038e-06,
      "loss": 0.4005,
      "step": 30000
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7155584528465885e-06,
      "loss": 0.405,
      "step": 30500
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.5797479356801393e-06,
      "loss": 0.3949,
      "step": 31000
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.4439374185136898e-06,
      "loss": 0.393,
      "step": 31500
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.3081269013472406e-06,
      "loss": 0.3974,
      "step": 32000
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.172316384180791e-06,
      "loss": 0.4052,
      "step": 32500
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.0365058670143417e-06,
      "loss": 0.3938,
      "step": 33000
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.006953498478922e-07,
      "loss": 0.3958,
      "step": 33500
    },
    {
      "epoch": 2.77,
      "learning_rate": 7.648848326814429e-07,
      "loss": 0.3917,
      "step": 34000
    },
    {
      "epoch": 2.81,
      "learning_rate": 6.290743155149935e-07,
      "loss": 0.4079,
      "step": 34500
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.932637983485442e-07,
      "loss": 0.3902,
      "step": 35000
    },
    {
      "epoch": 2.89,
      "learning_rate": 3.574532811820948e-07,
      "loss": 0.3802,
      "step": 35500
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.216427640156454e-07,
      "loss": 0.3934,
      "step": 36000
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.583224684919601e-08,
      "loss": 0.4002,
      "step": 36500
    },
    {
      "epoch": 3.0,
      "step": 36816,
      "total_flos": 1275249174174720.0,
      "train_runtime": 10484.4187,
      "train_samples_per_second": 3.511
    }
  ],
  "max_steps": 36816,
  "num_train_epochs": 3,
  "total_flos": 1275249174174720.0,
  "trial_name": null,
  "trial_params": null
}