{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 26990,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.9073731011485735e-05,
      "loss": 1.4116,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8147462022971475e-05,
      "loss": 1.4254,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.722119303445721e-05,
      "loss": 1.4358,
      "step": 1500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.629492404594295e-05,
      "loss": 1.4448,
      "step": 2000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.536865505742868e-05,
      "loss": 1.4453,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.444238606891441e-05,
      "loss": 1.4351,
      "step": 3000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.3516117080400145e-05,
      "loss": 1.4568,
      "step": 3500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.2589848091885885e-05,
      "loss": 1.4584,
      "step": 4000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.166357910337162e-05,
      "loss": 1.4641,
      "step": 4500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.073731011485736e-05,
      "loss": 1.4731,
      "step": 5000
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.981104112634309e-05,
      "loss": 1.4893,
      "step": 5500
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.888477213782883e-05,
      "loss": 1.4762,
      "step": 6000
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.795850314931456e-05,
      "loss": 1.4931,
      "step": 6500
    },
    {
      "epoch": 0.52,
      "learning_rate": 3.70322341608003e-05,
      "loss": 1.5,
      "step": 7000
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.6105965172286034e-05,
      "loss": 1.493,
      "step": 7500
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.5179696183771774e-05,
      "loss": 1.4961,
      "step": 8000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.425342719525751e-05,
      "loss": 1.5016,
      "step": 8500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.332715820674324e-05,
      "loss": 1.5055,
      "step": 9000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.240088921822897e-05,
      "loss": 1.5153,
      "step": 9500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.147462022971471e-05,
      "loss": 1.5185,
      "step": 10000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.0548351241200445e-05,
      "loss": 1.527,
      "step": 10500
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.9622082252686184e-05,
      "loss": 1.5339,
      "step": 11000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.8695813264171917e-05,
      "loss": 1.5337,
      "step": 11500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.776954427565765e-05,
      "loss": 1.5377,
      "step": 12000
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.684327528714339e-05,
      "loss": 1.5431,
      "step": 12500
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.5917006298629122e-05,
      "loss": 1.5574,
      "step": 13000
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.4990737310114858e-05,
      "loss": 1.5592,
      "step": 13500
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.4064468321600594e-05,
      "loss": 1.3231,
      "step": 14000
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.313819933308633e-05,
      "loss": 1.3354,
      "step": 14500
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.2211930344572066e-05,
      "loss": 1.3405,
      "step": 15000
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.1285661356057803e-05,
      "loss": 1.3607,
      "step": 15500
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.0359392367543535e-05,
      "loss": 1.3663,
      "step": 16000
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.943312337902927e-05,
      "loss": 1.3626,
      "step": 16500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8506854390515008e-05,
      "loss": 1.3191,
      "step": 17000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.7580585402000744e-05,
      "loss": 1.3267,
      "step": 17500
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.6654316413486477e-05,
      "loss": 1.3336,
      "step": 18000
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.5728047424972213e-05,
      "loss": 1.3735,
      "step": 18500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.4801778436457947e-05,
      "loss": 1.3832,
      "step": 19000
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.3875509447943683e-05,
      "loss": 1.389,
      "step": 19500
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.2949240459429418e-05,
      "loss": 1.4002,
      "step": 20000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.2022971470915156e-05,
      "loss": 1.4074,
      "step": 20500
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.109670248240089e-05,
      "loss": 1.4229,
      "step": 21000
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.0170433493886624e-05,
      "loss": 1.4368,
      "step": 21500
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.24416450537236e-06,
      "loss": 1.4427,
      "step": 22000
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.317895516858095e-06,
      "loss": 1.4402,
      "step": 22500
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.391626528343831e-06,
      "loss": 1.4554,
      "step": 23000
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.4653575398295665e-06,
      "loss": 1.4611,
      "step": 23500
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.539088551315303e-06,
      "loss": 1.4801,
      "step": 24000
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.612819562801038e-06,
      "loss": 1.4792,
      "step": 24500
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.686550574286773e-06,
      "loss": 1.4974,
      "step": 25000
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.7602815857725085e-06,
      "loss": 1.4973,
      "step": 25500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.8340125972582438e-06,
      "loss": 1.5153,
      "step": 26000
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.077436087439793e-07,
      "loss": 1.5293,
      "step": 26500
    },
    {
      "epoch": 2.0,
      "step": 26990,
      "total_flos": 2.5064145824199475e+17,
      "train_loss": 0.05643905973204775,
      "train_runtime": 943.9109,
      "train_samples_per_second": 285.921,
      "train_steps_per_second": 28.594
    }
  ],
  "max_steps": 26990,
  "num_train_epochs": 2,
  "total_flos": 2.5064145824199475e+17,
  "trial_name": null,
  "trial_params": null
}