run-32/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7574692442882249,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-32/checkpoint-642",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 642,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 7.148311614990234,
      "learning_rate": 0.00021207436837213607,
      "loss": 0.5606,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7129466900995899,
      "eval_f1": 0.5040485829959513,
      "eval_loss": 0.5154260396957397,
      "eval_mcc": 0.3157140452463625,
      "eval_precision": 0.594272076372315,
      "eval_recall": 0.437609841827768,
      "eval_runtime": 3.126,
      "eval_samples_per_second": 546.06,
      "eval_steps_per_second": 17.274,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.1550638675689697,
      "learning_rate": 0.0001817780300332595,
      "loss": 0.4995,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7393087287639133,
      "eval_f1": 0.5071982281284606,
      "eval_loss": 0.4903852641582489,
      "eval_mcc": 0.36859649869532995,
      "eval_precision": 0.6856287425149701,
      "eval_recall": 0.4024604569420035,
      "eval_runtime": 3.1521,
      "eval_samples_per_second": 541.536,
      "eval_steps_per_second": 17.131,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.5803661346435547,
      "learning_rate": 0.00015148169169438292,
      "loss": 0.479,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7574692442882249,
      "eval_f1": 0.541019955654102,
      "eval_loss": 0.48610907793045044,
      "eval_mcc": 0.41710213681200975,
      "eval_precision": 0.7327327327327328,
      "eval_recall": 0.4288224956063269,
      "eval_runtime": 3.1361,
      "eval_samples_per_second": 544.307,
      "eval_steps_per_second": 17.219,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 1712,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9251938490023766,
    "learning_rate": 0.00024237070671101265,
    "num_train_epochs": 8,
    "temperature": 22
  }
}