run-42/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7352079671939075,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-42/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": 8.560860633850098,
      "learning_rate": 0.00010512648397588192,
      "loss": 0.5768,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7106033977738723,
      "eval_f1": 0.5425925925925925,
      "eval_loss": 0.5306685566902161,
      "eval_mcc": 0.3328562997342372,
      "eval_precision": 0.5733855185909981,
      "eval_recall": 0.5149384885764499,
      "eval_runtime": 3.1739,
      "eval_samples_per_second": 537.821,
      "eval_steps_per_second": 17.014,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.4706881046295166,
      "learning_rate": 5.256324198794096e-05,
      "loss": 0.5157,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7328646748681898,
      "eval_f1": 0.5673624288425047,
      "eval_loss": 0.5048626661300659,
      "eval_mcc": 0.3784214504160885,
      "eval_precision": 0.6164948453608248,
      "eval_recall": 0.5254833040421792,
      "eval_runtime": 3.872,
      "eval_samples_per_second": 440.859,
      "eval_steps_per_second": 13.946,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.145559310913086,
      "learning_rate": 0.0,
      "loss": 0.4973,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7352079671939075,
      "eval_f1": 0.5461847389558233,
      "eval_loss": 0.4973718822002411,
      "eval_mcc": 0.3720623126055074,
      "eval_precision": 0.6370023419203747,
      "eval_recall": 0.4780316344463972,
      "eval_runtime": 3.1834,
      "eval_samples_per_second": 536.222,
      "eval_steps_per_second": 16.963,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 642,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9383404566107884,
    "learning_rate": 0.00015768972596382288,
    "num_train_epochs": 3,
    "temperature": 14
  }
}