{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 13260,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 4.811463046757165e-05,
      "loss": 0.1434,
      "step": 500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.622926093514329e-05,
      "loss": 0.0647,
      "step": 1000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.434389140271493e-05,
      "loss": 0.0529,
      "step": 1500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2458521870286574e-05,
      "loss": 0.049,
      "step": 2000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.057315233785822e-05,
      "loss": 0.0462,
      "step": 2500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.868778280542987e-05,
      "loss": 0.043,
      "step": 3000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.680241327300151e-05,
      "loss": 0.0418,
      "step": 3500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.491704374057315e-05,
      "loss": 0.0414,
      "step": 4000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3031674208144794e-05,
      "loss": 0.0375,
      "step": 4500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.114630467571644e-05,
      "loss": 0.0337,
      "step": 5000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9260935143288087e-05,
      "loss": 0.0322,
      "step": 5500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.737556561085973e-05,
      "loss": 0.0312,
      "step": 6000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5490196078431373e-05,
      "loss": 0.0313,
      "step": 6500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3604826546003017e-05,
      "loss": 0.031,
      "step": 7000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1719457013574662e-05,
      "loss": 0.0291,
      "step": 7500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9834087481146303e-05,
      "loss": 0.0286,
      "step": 8000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.794871794871795e-05,
      "loss": 0.0285,
      "step": 8500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6063348416289596e-05,
      "loss": 0.0284,
      "step": 9000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4177978883861237e-05,
      "loss": 0.0246,
      "step": 9500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.229260935143288e-05,
      "loss": 0.0236,
      "step": 10000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0407239819004526e-05,
      "loss": 0.0223,
      "step": 10500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.52187028657617e-06,
      "loss": 0.0233,
      "step": 11000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.636500754147813e-06,
      "loss": 0.0227,
      "step": 11500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.751131221719457e-06,
      "loss": 0.0239,
      "step": 12000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.865761689291101e-06,
      "loss": 0.0224,
      "step": 12500
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.80392156862745e-07,
      "loss": 0.0222,
      "step": 13000
    },
    {
      "epoch": 3.0,
      "step": 13260,
      "total_flos": 5955324393133404.0,
      "train_loss": 0.03731554852711489,
      "train_runtime": 2667.8261,
      "train_samples_per_second": 636.191,
      "train_steps_per_second": 4.97
    }
  ],
  "max_steps": 13260,
  "num_train_epochs": 3,
  "total_flos": 5955324393133404.0,
  "trial_name": null,
  "trial_params": null
}