run-15/checkpoint-576/trainer_state.json

{
  "best_metric": 0.8082191780821918,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-15/checkpoint-384",
  "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": 3.1954071521759033,
      "learning_rate": 4.7129202867976326e-05,
      "loss": 0.5731,
      "step": 192
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7896281800391389,
      "eval_f1": 0.8088888888888889,
      "eval_loss": 0.47491252422332764,
      "eval_mcc": 0.5913947215491926,
      "eval_precision": 0.741042345276873,
      "eval_recall": 0.8904109589041096,
      "eval_runtime": 66.6851,
      "eval_samples_per_second": 15.326,
      "eval_steps_per_second": 0.48,
      "step": 192
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.988830089569092,
      "learning_rate": 2.3564601433988163e-05,
      "loss": 0.4767,
      "step": 384
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8082191780821918,
      "eval_f1": 0.8208409506398537,
      "eval_loss": 0.4437917172908783,
      "eval_mcc": 0.6226500342007837,
      "eval_precision": 0.7701543739279588,
      "eval_recall": 0.8786692759295499,
      "eval_runtime": 66.9186,
      "eval_samples_per_second": 15.272,
      "eval_steps_per_second": 0.478,
      "step": 384
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.454801082611084,
      "learning_rate": 0.0,
      "loss": 0.4554,
      "step": 576
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8072407045009785,
      "eval_f1": 0.8233183856502243,
      "eval_loss": 0.4398359954357147,
      "eval_mcc": 0.624918005331598,
      "eval_precision": 0.7599337748344371,
      "eval_recall": 0.898238747553816,
      "eval_runtime": 66.52,
      "eval_samples_per_second": 15.364,
      "eval_steps_per_second": 0.481,
      "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": 0.8612888560371268,
    "learning_rate": 7.069380430196449e-05,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 16,
    "temperature": 23
  }
}