trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 666.6666666666666,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 166.66666666666666,
      "grad_norm": 0.11445357650518417,
      "learning_rate": 4.5025000000000003e-05,
      "loss": 0.0718,
      "step": 500
    },
    {
      "epoch": 333.3333333333333,
      "grad_norm": 0.11178029328584671,
      "learning_rate": 3.0050000000000002e-05,
      "loss": 0.0311,
      "step": 1000
    },
    {
      "epoch": 500.0,
      "grad_norm": 0.3114462196826935,
      "learning_rate": 1.5075000000000002e-05,
      "loss": 0.0251,
      "step": 1500
    },
    {
      "epoch": 666.6666666666666,
      "grad_norm": 0.05932952091097832,
      "learning_rate": 1e-07,
      "loss": 0.0224,
      "step": 2000
    },
    {
      "epoch": 666.6666666666666,
      "step": 2000,
      "total_flos": 1.4631512923711734e+18,
      "train_loss": 0.037614585399627684,
      "train_runtime": 4890.0384,
      "train_samples_per_second": 3.272,
      "train_steps_per_second": 0.409
    }
  ],
  "logging_steps": 500,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 667,
  "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": 1.4631512923711734e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}