trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 2860,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.749234135667396,
      "grad_norm": 0.8859563138601917,
      "learning_rate": 8.223278112677121e-06,
      "loss": 0.4662,
      "step": 1000
    },
    {
      "epoch": 3.497155361050328,
      "grad_norm": 0.8589195035414678,
      "learning_rate": 2.5158716126030835e-06,
      "loss": 0.1906,
      "step": 2000
    },
    {
      "epoch": 5.0,
      "step": 2860,
      "total_flos": 814512143859712.0,
      "train_loss": 0.24655424931666234,
      "train_runtime": 34363.2363,
      "train_samples_per_second": 7.976,
      "train_steps_per_second": 0.083
    }
  ],
  "logging_steps": 1000,
  "max_steps": 2860,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 814512143859712.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}