run-1/checkpoint-384/trainer_state.json

{
  "best_metric": 0.5,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-1/checkpoint-96",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 384,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.4667714238166809,
      "learning_rate": 7.759069356217338e-06,
      "loss": 0.317,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5,
      "eval_f1": 0.0,
      "eval_loss": 0.26975956559181213,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 28.3197,
      "eval_samples_per_second": 36.088,
      "eval_steps_per_second": 1.13,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.5722590088844299,
      "learning_rate": 5.1727129041448925e-06,
      "loss": 0.267,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5,
      "eval_f1": 0.0,
      "eval_loss": 0.2646586298942566,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 29.2799,
      "eval_samples_per_second": 34.905,
      "eval_steps_per_second": 1.093,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.38027867674827576,
      "learning_rate": 2.5863564520724462e-06,
      "loss": 0.2634,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5,
      "eval_f1": 0.0,
      "eval_loss": 0.26031577587127686,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 28.4418,
      "eval_samples_per_second": 35.933,
      "eval_steps_per_second": 1.125,
      "step": 288
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.5131752490997314,
      "learning_rate": 0.0,
      "loss": 0.2603,
      "step": 384
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5,
      "eval_f1": 0.0,
      "eval_loss": 0.2587716281414032,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 28.5668,
      "eval_samples_per_second": 35.776,
      "eval_steps_per_second": 1.12,
      "step": 384
    }
  ],
  "logging_steps": 500,
  "max_steps": 384,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 942780789120.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.2995113868248127,
    "learning_rate": 1.0345425808289785e-05,
    "num_train_epochs": 4,
    "temperature": 18
  }
}