checkpoint-16/trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 16,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.25,
      "grad_norm": 5.432992637451049,
      "learning_rate": 1e-05,
      "loss": 0.5341,
      "mean_token_accuracy": 0.8456295132637024,
      "num_tokens": 39650.0,
      "step": 1
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6810743546900195,
      "learning_rate": 1e-05,
      "loss": 0.3016,
      "mean_token_accuracy": 0.9056024551391602,
      "num_tokens": 403007.0,
      "step": 10
    }
  ],
  "logging_steps": 10,
  "max_steps": 64,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 16,
  "save_steps": 16,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1421657702400.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}