trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 5700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 2.3085,
      "step": 1900
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.9560242891311646,
      "eval_runtime": 18.4661,
      "eval_samples_per_second": 269.196,
      "eval_steps_per_second": 2.112,
      "step": 1900
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 1.9363,
      "step": 3800
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.87827467918396,
      "eval_runtime": 4.2837,
      "eval_samples_per_second": 1160.433,
      "eval_steps_per_second": 9.104,
      "step": 3800
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 1.873,
      "step": 5700
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.8482314348220825,
      "eval_runtime": 4.2658,
      "eval_samples_per_second": 1165.318,
      "eval_steps_per_second": 9.143,
      "step": 5700
    }
  ],
  "max_steps": 5700,
  "num_train_epochs": 3,
  "total_flos": 1.1914279759872e+16,
  "trial_name": null,
  "trial_params": null
}