trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.991249392318911,
  "global_step": 448,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0,
      "loss": 2.4259,
      "step": 1
    },
    {
      "epoch": 1.99,
      "step": 448,
      "total_flos": 1883191577149440.0,
      "train_loss": 0.7372576573065349,
      "train_runtime": 28706.0932,
      "train_samples_per_second": 4.013,
      "train_steps_per_second": 0.016
    }
  ],
  "max_steps": 448,
  "num_train_epochs": 2,
  "total_flos": 1883191577149440.0,
  "trial_name": null,
  "trial_params": null
}