{
  "best_metric": 0.8062622309197651,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-9/checkpoint-288",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 288,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.0101799964904785,
      "learning_rate": 5.9842495749639135e-05,
      "loss": 0.6452,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7514677103718199,
      "eval_f1": 0.7806563039723661,
      "eval_loss": 0.5494381785392761,
      "eval_precision": 0.6986089644513137,
      "eval_recall": 0.8845401174168297,
      "eval_runtime": 28.605,
      "eval_samples_per_second": 35.728,
      "eval_steps_per_second": 1.119,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0949199199676514,
      "learning_rate": 3.989499716642609e-05,
      "loss": 0.5311,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7925636007827789,
      "eval_f1": 0.8069216757741349,
      "eval_loss": 0.47911426424980164,
      "eval_precision": 0.7546848381601363,
      "eval_recall": 0.8669275929549902,
      "eval_runtime": 28.6227,
      "eval_samples_per_second": 35.706,
      "eval_steps_per_second": 1.118,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.1236400604248047,
      "learning_rate": 1.9947498583213046e-05,
      "loss": 0.493,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8062622309197651,
      "eval_f1": 0.8176795580110496,
      "eval_loss": 0.4623647630214691,
      "eval_precision": 0.7721739130434783,
      "eval_recall": 0.8688845401174168,
      "eval_runtime": 28.5439,
      "eval_samples_per_second": 35.805,
      "eval_steps_per_second": 1.121,
      "step": 288
    }
  ],
  "logging_steps": 500,
  "max_steps": 384,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 707085591840.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9656943950307342,
    "learning_rate": 7.978999433285218e-05,
    "num_train_epochs": 4,
    "temperature": 18
  }
}