trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 548,
  "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": 1.4855,
      "step": 274
    },
    {
      "epoch": 1.0,
      "eval_exact_match": 72.56616532587152,
      "eval_f1": 76.19648264629018,
      "eval_runtime": 3.3219,
      "eval_samples_per_second": 2581.933,
      "eval_steps_per_second": 8.128,
      "step": 274
    },
    {
      "epoch": 2.0,
      "learning_rate": 3e-05,
      "loss": 0.7957,
      "step": 548
    },
    {
      "epoch": 2.0,
      "eval_exact_match": 73.06750612102134,
      "eval_f1": 76.60875533208507,
      "eval_runtime": 2.9408,
      "eval_samples_per_second": 2916.565,
      "eval_steps_per_second": 9.181,
      "step": 548
    }
  ],
  "max_steps": 2740,
  "num_train_epochs": 10,
  "total_flos": 66076463333376.0,
  "trial_name": null,
  "trial_params": null
}