{
  "best_metric": 0.6047325730323792,
  "best_model_checkpoint": "BERT-WMM/run-4/checkpoint-268",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 804,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7248826291079812,
      "eval_loss": 0.6610550880432129,
      "eval_runtime": 2.0768,
      "eval_samples_per_second": 1025.597,
      "eval_steps_per_second": 64.521,
      "step": 134
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.755868544600939,
      "eval_loss": 0.6047325730323792,
      "eval_runtime": 2.0636,
      "eval_samples_per_second": 1032.163,
      "eval_steps_per_second": 64.934,
      "step": 268
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7568075117370892,
      "eval_loss": 0.6309530138969421,
      "eval_runtime": 2.6635,
      "eval_samples_per_second": 799.702,
      "eval_steps_per_second": 50.31,
      "step": 402
    },
    {
      "epoch": 3.73,
      "grad_norm": 9.669575691223145,
      "learning_rate": 4.323409306826513e-06,
      "loss": 0.5385,
      "step": 500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7577464788732394,
      "eval_loss": 0.6676124334335327,
      "eval_runtime": 2.0995,
      "eval_samples_per_second": 1014.528,
      "eval_steps_per_second": 63.825,
      "step": 536
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7530516431924883,
      "eval_loss": 0.694180965423584,
      "eval_runtime": 2.1225,
      "eval_samples_per_second": 1003.547,
      "eval_steps_per_second": 63.134,
      "step": 670
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7502347417840376,
      "eval_loss": 0.7122268676757812,
      "eval_runtime": 2.0528,
      "eval_samples_per_second": 1037.598,
      "eval_steps_per_second": 65.276,
      "step": 804
    }
  ],
  "logging_steps": 500,
  "max_steps": 804,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 709335274032504.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.1434279877264858e-05,
    "num_train_epochs": 6,
    "per_device_train_batch_size": 64,
    "weight_decay": 0.007798418229484321
  }
}