trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9997457842555715,
  "eval_steps": 500,
  "global_step": 26550,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 6.667043314500941e-05,
      "loss": 2.8102,
      "step": 8850
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.3348399246704334e-05,
      "loss": 2.5375,
      "step": 17701
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.0131826741996235e-08,
      "loss": 2.4492,
      "step": 26550
    }
  ],
  "logging_steps": 500,
  "max_steps": 26550,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}