checkpoint-10/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 5,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 21.60628890991211,
      "learning_rate": 2e-05,
      "loss": 1.2546,
      "step": 1
    },
    {
      "epoch": 2.0,
      "grad_norm": 21.794620513916016,
      "learning_rate": 1.9863613034027224e-05,
      "loss": 1.2546,
      "step": 2
    },
    {
      "epoch": 3.0,
      "grad_norm": 13.444245338439941,
      "learning_rate": 1.9458172417006347e-05,
      "loss": 1.1815,
      "step": 3
    },
    {
      "epoch": 4.0,
      "grad_norm": 9.568578720092773,
      "learning_rate": 1.879473751206489e-05,
      "loss": 1.0718,
      "step": 4
    },
    {
      "epoch": 5.0,
      "grad_norm": 8.503974914550781,
      "learning_rate": 1.789140509396394e-05,
      "loss": 0.9886,
      "step": 5
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9387755102040817,
      "eval_accuracy_<|content|>": 0.0,
      "eval_accuracy_<|from|>": -1,
      "eval_accuracy_<|recipient|>": -1,
      "eval_accuracy_<|stop|>": 0.0,
      "eval_accuracy_first_token": 1.0,
      "eval_accuracy_first_token_get": 1.0,
      "eval_accuracy_first_token_get_total": 2,
      "eval_accuracy_total_num_<|content|>": 2,
      "eval_accuracy_total_num_<|from|>": 0,
      "eval_accuracy_total_num_<|recipient|>": 0,
      "eval_accuracy_total_num_<|stop|>": 2,
      "eval_first_token_param_values": 0.9,
      "eval_first_token_param_values_total": 10,
      "eval_loss": 0.897951602935791,
      "eval_perplexity": 1.0108013242189644,
      "eval_runtime": 0.885,
      "eval_samples_per_second": 1.13,
      "eval_steps_per_second": 1.13,
      "eval_total_number_first_token": 2,
      "step": 5
    },
    {
      "epoch": 6.0,
      "grad_norm": 8.138020515441895,
      "learning_rate": 1.6772815716257414e-05,
      "loss": 0.8967,
      "step": 6
    },
    {
      "epoch": 7.0,
      "grad_norm": 8.324158668518066,
      "learning_rate": 1.5469481581224274e-05,
      "loss": 0.8374,
      "step": 7
    },
    {
      "epoch": 8.0,
      "grad_norm": 8.041045188903809,
      "learning_rate": 1.4016954246529697e-05,
      "loss": 0.7753,
      "step": 8
    },
    {
      "epoch": 9.0,
      "grad_norm": 7.898632526397705,
      "learning_rate": 1.2454854871407993e-05,
      "loss": 0.7203,
      "step": 9
    },
    {
      "epoch": 10.0,
      "grad_norm": 6.87092924118042,
      "learning_rate": 1.0825793454723325e-05,
      "loss": 0.6677,
      "step": 10
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9489795918367347,
      "eval_accuracy_<|content|>": 0.0,
      "eval_accuracy_<|from|>": -1,
      "eval_accuracy_<|recipient|>": -1,
      "eval_accuracy_<|stop|>": 0.0,
      "eval_accuracy_first_token": 1.0,
      "eval_accuracy_first_token_get": 1.0,
      "eval_accuracy_first_token_get_total": 2,
      "eval_accuracy_total_num_<|content|>": 2,
      "eval_accuracy_total_num_<|from|>": 0,
      "eval_accuracy_total_num_<|recipient|>": 0,
      "eval_accuracy_total_num_<|stop|>": 2,
      "eval_first_token_param_values": 0.9,
      "eval_first_token_param_values_total": 10,
      "eval_loss": 0.627693235874176,
      "eval_perplexity": 1.0075382125074093,
      "eval_runtime": 0.878,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 1.139,
      "eval_total_number_first_token": 2,
      "step": 10
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 5,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3967665957765120.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}