trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.97196261682243,
  "eval_steps": 500,
  "global_step": 159,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.97,
      "step": 159,
      "total_flos": 4.448210991552922e+16,
      "train_loss": 2.5530455007493122,
      "train_runtime": 302.3975,
      "train_samples_per_second": 8.433,
      "train_steps_per_second": 0.526
    }
  ],
  "logging_steps": 500,
  "max_steps": 159,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4.448210991552922e+16,
  "trial_name": null,
  "trial_params": null
}