trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 1565,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.6,
      "learning_rate": 3.402555910543131e-05,
      "loss": 0.8327,
      "step": 500
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.805111821086262e-05,
      "loss": 0.0413,
      "step": 1000
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.0766773162939296e-06,
      "loss": 0.0192,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "step": 1565,
      "total_flos": 42196507361280.0,
      "train_loss": 0.2860380729547324,
      "train_runtime": 110.266,
      "train_samples_per_second": 56.681,
      "train_steps_per_second": 14.193
    }
  ],
  "max_steps": 1565,
  "num_train_epochs": 5,
  "total_flos": 42196507361280.0,
  "trial_name": null,
  "trial_params": null
}