trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 66.66666666666667,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 33.33,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 2.7852,
      "step": 100
    },
    {
      "epoch": 66.67,
      "learning_rate": 0.0,
      "loss": 0.1412,
      "step": 200
    },
    {
      "epoch": 66.67,
      "step": 200,
      "total_flos": 368165088860112.0,
      "train_loss": 1.4632204246520997,
      "train_runtime": 100.8189,
      "train_samples_per_second": 23.805,
      "train_steps_per_second": 1.984
    }
  ],
  "max_steps": 200,
  "num_train_epochs": 67,
  "total_flos": 368165088860112.0,
  "trial_name": null,
  "trial_params": null
}