run-20/checkpoint-576/trainer_state.json

{
  "best_metric": 0.8258317025440313,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-20/checkpoint-576",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 576,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 11.2067289352417,
      "learning_rate": 0.00015355053489920167,
      "loss": 0.5758,
      "step": 192
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7906066536203522,
      "eval_f1": 0.7938342967244701,
      "eval_loss": 0.4803670346736908,
      "eval_mcc": 0.5814984242947744,
      "eval_precision": 0.7817836812144212,
      "eval_recall": 0.8062622309197651,
      "eval_runtime": 66.3514,
      "eval_samples_per_second": 15.403,
      "eval_steps_per_second": 0.482,
      "step": 192
    },
    {
      "epoch": 2.0,
      "grad_norm": 13.654583930969238,
      "learning_rate": 7.677526744960083e-05,
      "loss": 0.4629,
      "step": 384
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8160469667318982,
      "eval_f1": 0.824953445065177,
      "eval_loss": 0.421450674533844,
      "eval_mcc": 0.6353923506365322,
      "eval_precision": 0.7868561278863233,
      "eval_recall": 0.8669275929549902,
      "eval_runtime": 67.0427,
      "eval_samples_per_second": 15.244,
      "eval_steps_per_second": 0.477,
      "step": 384
    },
    {
      "epoch": 3.0,
      "grad_norm": 6.51780891418457,
      "learning_rate": 0.0,
      "loss": 0.4179,
      "step": 576
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8258317025440313,
      "eval_f1": 0.836697247706422,
      "eval_loss": 0.4053334593772888,
      "eval_mcc": 0.6575110994179842,
      "eval_precision": 0.7875647668393783,
      "eval_recall": 0.8923679060665362,
      "eval_runtime": 67.0931,
      "eval_samples_per_second": 15.233,
      "eval_steps_per_second": 0.477,
      "step": 576
    }
  ],
  "logging_steps": 500,
  "max_steps": 576,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 707085591840.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9932487320274002,
    "learning_rate": 0.0002303258023488025,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 16,
    "temperature": 12
  }
}