{
  "best_metric": 0.7821100917431193,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-2/checkpoint-527",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 527,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 3.8070333003997803,
      "learning_rate": 3.0759690838398e-05,
      "loss": 0.8585,
      "step": 527
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7821100917431193,
      "eval_loss": 0.7264055609703064,
      "eval_runtime": 2.6688,
      "eval_samples_per_second": 326.736,
      "eval_steps_per_second": 2.623,
      "step": 527
    }
  ],
  "logging_steps": 500,
  "max_steps": 5270,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 8057768762220.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9025220801031371,
    "learning_rate": 3.4177434264886666e-05,
    "num_train_epochs": 10,
    "temperature": 12
  }
}