trainer_state.json

{
  "best_metric": 0.4001782238483429,
  "best_model_checkpoint": "./platzi-distilroberta-base-mrpc-glue-wgcv/checkpoint-500",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 690,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.1739130434782608,
      "grad_norm": 3.1889848709106445,
      "learning_rate": 1.3768115942028985e-05,
      "loss": 0.409,
      "step": 500
    },
    {
      "epoch": 2.1739130434782608,
      "eval_accuracy": 0.8455882352941176,
      "eval_f1": 0.8835489833641405,
      "eval_loss": 0.4001782238483429,
      "eval_runtime": 54.0627,
      "eval_samples_per_second": 7.547,
      "eval_steps_per_second": 0.481,
      "step": 500
    },
    {
      "epoch": 3.0,
      "step": 690,
      "total_flos": 216973319786112.0,
      "train_loss": 0.34854334955630095,
      "train_runtime": 5190.8247,
      "train_samples_per_second": 2.12,
      "train_steps_per_second": 0.133
    }
  ],
  "logging_steps": 500,
  "max_steps": 690,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 216973319786112.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}