trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.764705882352941,
  "eval_steps": 500,
  "global_step": 8,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.858464241027832,
      "learning_rate": 0.00015000000000000001,
      "loss": 9.5253,
      "step": 2
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 1.1875278949737549,
      "learning_rate": 0.0001,
      "loss": 9.2917,
      "step": 4
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 1.371861219406128,
      "learning_rate": 5e-05,
      "loss": 9.0606,
      "step": 6
    },
    {
      "epoch": 3.764705882352941,
      "grad_norm": 1.4962043762207031,
      "learning_rate": 0.0,
      "loss": 8.9123,
      "step": 8
    },
    {
      "epoch": 3.764705882352941,
      "step": 8,
      "total_flos": 36858669560664.0,
      "train_loss": 9.197454690933228,
      "train_runtime": 36.5767,
      "train_samples_per_second": 3.609,
      "train_steps_per_second": 0.219
    }
  ],
  "logging_steps": 2,
  "max_steps": 8,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 36858669560664.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}