trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 14.434160583109545,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 1.1748,
      "step": 10
    },
    {
      "epoch": 2.0,
      "grad_norm": 72.2505883022695,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.6402,
      "step": 20
    },
    {
      "epoch": 3.0,
      "grad_norm": 9.760732101520077,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.2264,
      "step": 30
    },
    {
      "epoch": 4.0,
      "grad_norm": 5.553741412756681,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.08,
      "step": 40
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.2011073393718297,
      "learning_rate": 0.0,
      "loss": 0.0332,
      "step": 50
    },
    {
      "epoch": 5.0,
      "step": 50,
      "total_flos": 350721146880.0,
      "train_loss": 0.43092278122901917,
      "train_runtime": 122.9688,
      "train_samples_per_second": 6.099,
      "train_steps_per_second": 0.407
    }
  ],
  "logging_steps": 10,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 350721146880.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}