trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9900687547746372,
  "eval_steps": 500,
  "global_step": 81,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.24446142093200918,
      "grad_norm": 1.040741205215454,
      "learning_rate": 9.435054165891109e-05,
      "loss": 10.5551,
      "step": 20
    },
    {
      "epoch": 0.48892284186401835,
      "grad_norm": 0.43350982666015625,
      "learning_rate": 6.0821980696905146e-05,
      "loss": 8.4119,
      "step": 40
    },
    {
      "epoch": 0.7333842627960275,
      "grad_norm": 0.21506506204605103,
      "learning_rate": 1.9561928549563968e-05,
      "loss": 7.8467,
      "step": 60
    },
    {
      "epoch": 0.9778456837280367,
      "grad_norm": 0.206473708152771,
      "learning_rate": 4.7588920907110094e-08,
      "loss": 7.7607,
      "step": 80
    }
  ],
  "logging_steps": 20,
  "max_steps": 81,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.834302370324152e+17,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": null
}