trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.001632386549134835,
  "eval_steps": 5,
  "global_step": 5,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.9185667752443e-05,
      "loss": 1.1026,
      "step": 5
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.1494640111923218,
      "eval_runtime": 18.2071,
      "eval_samples_per_second": 109.847,
      "eval_steps_per_second": 1.758,
      "step": 5
    }
  ],
  "logging_steps": 5,
  "max_steps": 307,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5,
  "total_flos": 10451681280000.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}