{
  "best_metric": 0.6369415521621704,
  "best_model_checkpoint": "BERT-WMM/run-0/checkpoint-533",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 2132,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.94,
      "grad_norm": 11.681629180908203,
      "learning_rate": 2.5842410540171958e-05,
      "loss": 0.7181,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7525821596244131,
      "eval_loss": 0.6369415521621704,
      "eval_runtime": 1.9836,
      "eval_samples_per_second": 1073.792,
      "eval_steps_per_second": 67.553,
      "step": 533
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.492458820343018,
      "learning_rate": 2.1053231418568556e-05,
      "loss": 0.4571,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7497652582159624,
      "eval_loss": 0.6769182682037354,
      "eval_runtime": 2.0259,
      "eval_samples_per_second": 1051.385,
      "eval_steps_per_second": 66.144,
      "step": 1066
    },
    {
      "epoch": 2.81,
      "grad_norm": 13.743987083435059,
      "learning_rate": 1.626405229696515e-05,
      "loss": 0.2889,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7596244131455399,
      "eval_loss": 0.8726277351379395,
      "eval_runtime": 2.0636,
      "eval_samples_per_second": 1032.174,
      "eval_steps_per_second": 64.935,
      "step": 1599
    },
    {
      "epoch": 3.75,
      "grad_norm": 31.317466735839844,
      "learning_rate": 1.147487317536175e-05,
      "loss": 0.1815,
      "step": 2000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7572769953051643,
      "eval_loss": 1.311642050743103,
      "eval_runtime": 2.081,
      "eval_samples_per_second": 1023.549,
      "eval_steps_per_second": 64.392,
      "step": 2132
    }
  ],
  "logging_steps": 500,
  "max_steps": 3198,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 676890100708920.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 3.063158966177536e-05,
    "num_train_epochs": 6,
    "per_device_train_batch_size": 16,
    "weight_decay": 2.6426532465921284e-05
  }
}