last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.3333333333333335,
  "eval_steps": 200,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.33,
      "learning_rate": 0.0002,
      "loss": 0.1441,
      "step": 200
    },
    {
      "epoch": 3.33,
      "eval_runtime": 23.4067,
      "eval_samples_per_second": 4.272,
      "eval_steps_per_second": 0.171,
      "step": 200
    }
  ],
  "logging_steps": 200,
  "max_steps": 960,
  "num_train_epochs": 16,
  "save_steps": 200,
  "total_flos": 1.27974621315072e+17,
  "trial_name": null,
  "trial_params": null
}