last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.00014873261222805173,
  "eval_steps": 2000,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 8.451894760131836,
      "learning_rate": 1.9999999959757473e-05,
      "loss": 1.835,
      "step": 200
    }
  ],
  "logging_steps": 200,
  "max_steps": 6723475,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 200,
  "total_flos": 2438121270509568.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}