trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 24,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.33834290504455566,
      "learning_rate": 8.966766701456177e-05,
      "loss": 0.5297,
      "num_input_tokens_seen": 81920,
      "step": 5
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.35304397344589233,
      "learning_rate": 6.294095225512603e-05,
      "loss": 0.4377,
      "num_input_tokens_seen": 159872,
      "step": 10
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.21805448830127716,
      "learning_rate": 3.086582838174551e-05,
      "loss": 0.3134,
      "num_input_tokens_seen": 241792,
      "step": 15
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.21976211667060852,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.3175,
      "num_input_tokens_seen": 323712,
      "step": 20
    },
    {
      "epoch": 2.0,
      "num_input_tokens_seen": 389248,
      "step": 24,
      "total_flos": 9164234264936448.0,
      "train_loss": 0.3852280577023824,
      "train_runtime": 121.43,
      "train_samples_per_second": 2.965,
      "train_steps_per_second": 0.198
    }
  ],
  "logging_steps": 5,
  "max_steps": 24,
  "num_input_tokens_seen": 389248,
  "num_train_epochs": 2,
  "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": 9164234264936448.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}