run-17/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7076742823667252,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-17/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": 5.922779083251953,
      "learning_rate": 6.801673821225247e-05,
      "loss": 0.5484,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6906854130052724,
      "eval_f1": 0.4741035856573705,
      "eval_loss": 0.5109518766403198,
      "eval_mcc": 0.2652167942468068,
      "eval_precision": 0.5471264367816092,
      "eval_recall": 0.4182776801405975,
      "eval_runtime": 3.1422,
      "eval_samples_per_second": 543.246,
      "eval_steps_per_second": 17.185,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.857205629348755,
      "learning_rate": 5.4413390569801966e-05,
      "loss": 0.5052,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7070884592852958,
      "eval_f1": 0.50199203187251,
      "eval_loss": 0.49773725867271423,
      "eval_mcc": 0.30514190305815403,
      "eval_precision": 0.5793103448275863,
      "eval_recall": 0.4428822495606327,
      "eval_runtime": 3.1244,
      "eval_samples_per_second": 546.344,
      "eval_steps_per_second": 17.283,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.345863103866577,
      "learning_rate": 4.0810042927351476e-05,
      "loss": 0.491,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7076742823667252,
      "eval_f1": 0.48503611971104243,
      "eval_loss": 0.490793377161026,
      "eval_mcc": 0.29827513576932563,
      "eval_precision": 0.5875,
      "eval_recall": 0.4130052724077329,
      "eval_runtime": 3.144,
      "eval_samples_per_second": 542.939,
      "eval_steps_per_second": 17.176,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 1284,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.8734730758781912,
    "learning_rate": 8.162008585470295e-05,
    "num_train_epochs": 6,
    "temperature": 22
  }
}