trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9761904761904763,
  "global_step": 2125,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.7002801120448179,
      "learning_rate": 3.832866479925304e-05,
      "loss": 0.2923659973144531,
      "step": 500
    },
    {
      "epoch": 1.4005602240896358,
      "learning_rate": 2.665732959850607e-05,
      "loss": 0.12518920135498046,
      "step": 1000
    },
    {
      "epoch": 2.100840336134454,
      "learning_rate": 1.4985994397759103e-05,
      "loss": 0.08236197662353516,
      "step": 1500
    },
    {
      "epoch": 2.8011204481792715,
      "learning_rate": 3.3146591970121383e-06,
      "loss": 0.03941415023803711,
      "step": 2000
    }
  ],
  "max_steps": 2142,
  "num_train_epochs": 3,
  "total_flos": 2843017475953152,
  "trial_name": null,
  "trial_params": null
}