run-33/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7141183362624487,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-33/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": 6.1991143226623535,
      "learning_rate": 6.724081136915917e-05,
      "loss": 0.5853,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6924428822495606,
      "eval_f1": 0.46373850868232896,
      "eval_loss": 0.5429773330688477,
      "eval_mcc": 0.26278004135127175,
      "eval_precision": 0.5536585365853659,
      "eval_recall": 0.3989455184534271,
      "eval_runtime": 3.1406,
      "eval_samples_per_second": 543.53,
      "eval_steps_per_second": 17.194,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.212712287902832,
      "learning_rate": 5.883570994801428e-05,
      "loss": 0.5367,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7070884592852958,
      "eval_f1": 0.536178107606679,
      "eval_loss": 0.5276417136192322,
      "eval_mcc": 0.32417590252227235,
      "eval_precision": 0.5677799607072691,
      "eval_recall": 0.507908611599297,
      "eval_runtime": 3.1564,
      "eval_samples_per_second": 540.802,
      "eval_steps_per_second": 17.108,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.126155376434326,
      "learning_rate": 5.043060852686938e-05,
      "loss": 0.5195,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7141183362624487,
      "eval_f1": 0.5080645161290324,
      "eval_loss": 0.519816517829895,
      "eval_mcc": 0.31950417692011085,
      "eval_precision": 0.5957446808510638,
      "eval_recall": 0.4428822495606327,
      "eval_runtime": 3.148,
      "eval_samples_per_second": 542.251,
      "eval_steps_per_second": 17.154,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 1926,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9360715938313949,
    "learning_rate": 7.564591279030407e-05,
    "num_train_epochs": 9,
    "temperature": 28
  }
}