trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 216,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 3e-05,
      "loss": 0.9073,
      "step": 108
    },
    {
      "epoch": 1.0,
      "eval_exact_match": 72.7060743849831,
      "eval_f1": 76.14881023806663,
      "eval_runtime": 8.3995,
      "eval_samples_per_second": 1021.136,
      "eval_steps_per_second": 3.214,
      "step": 108
    },
    {
      "epoch": 2.0,
      "learning_rate": 3e-05,
      "loss": 0.4488,
      "step": 216
    },
    {
      "epoch": 2.0,
      "eval_exact_match": 73.13746065057713,
      "eval_f1": 76.44856070179605,
      "eval_runtime": 7.6164,
      "eval_samples_per_second": 1126.117,
      "eval_steps_per_second": 3.545,
      "step": 216
    }
  ],
  "max_steps": 2160,
  "num_train_epochs": 20,
  "total_flos": 67913918709760.0,
  "trial_name": null,
  "trial_params": null
}