{
  "best_metric": 0.898797983714618,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-1/checkpoint-1935",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1935,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8288795948028564,
      "learning_rate": 0.0001908740631878717,
      "loss": 0.3032,
      "step": 645
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8892981775882125,
      "eval_f1": 0.12557427258805512,
      "eval_loss": 0.29040220379829407,
      "eval_mcc": 0.15515682588139135,
      "eval_precision": 0.4880952380952381,
      "eval_recall": 0.07205623901581722,
      "eval_runtime": 9.95,
      "eval_samples_per_second": 518.391,
      "eval_steps_per_second": 16.281,
      "step": 645
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.593945324420929,
      "learning_rate": 9.543703159393586e-05,
      "loss": 0.2874,
      "step": 1290
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8964715005816207,
      "eval_f1": 0.16300940438871472,
      "eval_loss": 0.2867793142795563,
      "eval_mcc": 0.23910805095304313,
      "eval_precision": 0.7536231884057971,
      "eval_recall": 0.0913884007029877,
      "eval_runtime": 9.9641,
      "eval_samples_per_second": 517.657,
      "eval_steps_per_second": 16.258,
      "step": 1290
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2576031684875488,
      "learning_rate": 0.0,
      "loss": 0.2765,
      "step": 1935
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.898797983714618,
      "eval_f1": 0.24347826086956523,
      "eval_loss": 0.2753896117210388,
      "eval_mcc": 0.2888763289104669,
      "eval_precision": 0.6942148760330579,
      "eval_recall": 0.14762741652021089,
      "eval_runtime": 9.9488,
      "eval_samples_per_second": 518.456,
      "eval_steps_per_second": 16.283,
      "step": 1935
    }
  ],
  "logging_steps": 500,
  "max_steps": 1935,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4760827597800.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.8638056037769629,
    "learning_rate": 0.0002863110947818076,
    "num_train_epochs": 3,
    "temperature": 11
  }
}