trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.777777777777779,
  "eval_steps": 500,
  "global_step": 550,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.89,
      "grad_norm": 1.137364387512207,
      "learning_rate": 3.3834586466165413e-06,
      "loss": 1.1857,
      "step": 500
    },
    {
      "epoch": 8.89,
      "eval_loss": 0.40131083130836487,
      "eval_runtime": 12.4257,
      "eval_samples_per_second": 36.215,
      "eval_steps_per_second": 2.334,
      "step": 500
    }
  ],
  "logging_steps": 500,
  "max_steps": 560,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 50,
  "total_flos": 1.194443308929024e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}