{
  "best_metric": 0.5207920792079208,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-2/checkpoint-594",
  "epoch": 9.0,
  "eval_steps": 500,
  "global_step": 2673,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.46941760182380676,
      "learning_rate": 1.0066413527942762e-05,
      "loss": 0.2155,
      "step": 297
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.497029702970297,
      "eval_loss": 0.18417517840862274,
      "eval_runtime": 14.6034,
      "eval_samples_per_second": 34.581,
      "eval_steps_per_second": 1.096,
      "step": 297
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3071568012237549,
      "learning_rate": 8.808111836949918e-06,
      "loss": 0.1831,
      "step": 594
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5207920792079208,
      "eval_loss": 0.17807520925998688,
      "eval_runtime": 14.6044,
      "eval_samples_per_second": 34.579,
      "eval_steps_per_second": 1.096,
      "step": 594
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.35216400027275085,
      "learning_rate": 7.549810145957073e-06,
      "loss": 0.1804,
      "step": 891
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5207920792079208,
      "eval_loss": 0.17647753655910492,
      "eval_runtime": 15.4786,
      "eval_samples_per_second": 32.626,
      "eval_steps_per_second": 1.034,
      "step": 891
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.26479315757751465,
      "learning_rate": 6.291508454964228e-06,
      "loss": 0.179,
      "step": 1188
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.1757788509130478,
      "eval_runtime": 13.9632,
      "eval_samples_per_second": 36.166,
      "eval_steps_per_second": 1.146,
      "step": 1188
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.33599618077278137,
      "learning_rate": 5.033206763971381e-06,
      "loss": 0.1781,
      "step": 1485
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.17515264451503754,
      "eval_runtime": 14.2891,
      "eval_samples_per_second": 35.342,
      "eval_steps_per_second": 1.12,
      "step": 1485
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.21752449870109558,
      "learning_rate": 3.7749050729785363e-06,
      "loss": 0.1781,
      "step": 1782
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.17485138773918152,
      "eval_runtime": 13.9491,
      "eval_samples_per_second": 36.203,
      "eval_steps_per_second": 1.147,
      "step": 1782
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.2491326630115509,
      "learning_rate": 2.5166033819856906e-06,
      "loss": 0.1777,
      "step": 2079
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.17470771074295044,
      "eval_runtime": 14.4075,
      "eval_samples_per_second": 35.051,
      "eval_steps_per_second": 1.111,
      "step": 2079
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.5971789956092834,
      "learning_rate": 1.2583016909928453e-06,
      "loss": 0.1776,
      "step": 2376
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.17442195117473602,
      "eval_runtime": 14.8484,
      "eval_samples_per_second": 34.011,
      "eval_steps_per_second": 1.078,
      "step": 2376
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.2817859947681427,
      "learning_rate": 0.0,
      "loss": 0.1773,
      "step": 2673
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.5168316831683168,
      "eval_loss": 0.17444071173667908,
      "eval_runtime": 13.7553,
      "eval_samples_per_second": 36.713,
      "eval_steps_per_second": 1.163,
      "step": 2673
    }
  ],
  "logging_steps": 500,
  "max_steps": 2673,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "total_flos": 7000589548260.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.20650329892275032,
    "learning_rate": 1.1324715218935609e-05,
    "num_train_epochs": 9,
    "temperature": 10
  }
}