trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 3516,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.85022,
      "eval_loss": 0.33765363693237305,
      "eval_runtime": 102.5233,
      "eval_samples_per_second": 487.694,
      "step": 1172
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.0053050397877986e-05,
      "loss": 0.38,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.86544,
      "eval_loss": 0.3076673150062561,
      "eval_runtime": 123.4792,
      "eval_samples_per_second": 404.926,
      "step": 2344
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.86948,
      "eval_loss": 0.31286707520484924,
      "eval_runtime": 123.9806,
      "eval_samples_per_second": 403.289,
      "step": 3516
    },
    {
      "epoch": 3.0,
      "step": 3516,
      "total_flos": 0,
      "train_runtime": 2882.6525,
      "train_samples_per_second": 1.22
    }
  ],
  "max_steps": 3516,
  "num_train_epochs": 3,
  "total_flos": 0,
  "trial_name": null,
  "trial_params": null
}