trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 26,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.77,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 0.5091,
      "step": 10
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.1673,
      "step": 20
    },
    {
      "epoch": 1.54,
      "eval_exact_match": 0.0,
      "eval_f1": 15.883333333333333,
      "step": 20
    },
    {
      "epoch": 2.0,
      "step": 26,
      "total_flos": 60973645824000.0,
      "train_loss": 0.28480225572219264,
      "train_runtime": 22.8013,
      "train_samples_per_second": 8.771,
      "train_steps_per_second": 1.14
    }
  ],
  "max_steps": 26,
  "num_train_epochs": 2,
  "total_flos": 60973645824000.0,
  "trial_name": null,
  "trial_params": null
}