{
  "best_metric": 0.8297455968688845,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-10/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": 1.9170680046081543,
      "learning_rate": 0.0005212663608648181,
      "loss": 0.5641,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.799412915851272,
      "eval_f1": 0.8197009674582234,
      "eval_loss": 0.47259676456451416,
      "eval_precision": 0.744408945686901,
      "eval_recall": 0.9119373776908023,
      "eval_runtime": 30.8424,
      "eval_samples_per_second": 33.136,
      "eval_steps_per_second": 1.038,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.4848690032958984,
      "learning_rate": 0.00026063318043240905,
      "loss": 0.456,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.824853228962818,
      "eval_f1": 0.8406055209260909,
      "eval_loss": 0.4349122643470764,
      "eval_precision": 0.7712418300653595,
      "eval_recall": 0.923679060665362,
      "eval_runtime": 31.1515,
      "eval_samples_per_second": 32.807,
      "eval_steps_per_second": 1.027,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.3911333084106445,
      "learning_rate": 0.0,
      "loss": 0.4063,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8297455968688845,
      "eval_f1": 0.8421052631578947,
      "eval_loss": 0.42057597637176514,
      "eval_precision": 0.7851099830795262,
      "eval_recall": 0.9080234833659491,
      "eval_runtime": 32.1366,
      "eval_samples_per_second": 31.802,
      "eval_steps_per_second": 0.996,
      "step": 288
    }
  ],
  "logging_steps": 500,
  "max_steps": 288,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 707085591840.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9610512963031814,
    "learning_rate": 0.0007818995412972273,
    "num_train_epochs": 3,
    "temperature": 24
  }
}