run-46/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7486818980667839,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-46/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": 4.412960052490234,
      "learning_rate": 0.00020694993120331544,
      "loss": 0.4902,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7164616285881664,
      "eval_f1": 0.39800995024875624,
      "eval_loss": 0.4596139192581177,
      "eval_mcc": 0.2945528643838721,
      "eval_precision": 0.6808510638297872,
      "eval_recall": 0.281195079086116,
      "eval_runtime": 3.1802,
      "eval_samples_per_second": 536.76,
      "eval_steps_per_second": 16.98,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3837729692459106,
      "learning_rate": 0.00018395549440294705,
      "loss": 0.4478,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7387229056824839,
      "eval_f1": 0.4520884520884521,
      "eval_loss": 0.44098830223083496,
      "eval_mcc": 0.3627159463012621,
      "eval_precision": 0.7510204081632653,
      "eval_recall": 0.3233743409490334,
      "eval_runtime": 4.0058,
      "eval_samples_per_second": 426.137,
      "eval_steps_per_second": 13.481,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0605193376541138,
      "learning_rate": 0.00016096105760257865,
      "loss": 0.4298,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7486818980667839,
      "eval_f1": 0.48,
      "eval_loss": 0.4385921359062195,
      "eval_mcc": 0.39214620237686104,
      "eval_precision": 0.7734375,
      "eval_recall": 0.34797891036906853,
      "eval_runtime": 3.1306,
      "eval_samples_per_second": 545.257,
      "eval_steps_per_second": 17.249,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 2140,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.7874456347292355,
    "learning_rate": 0.0002299443680036838,
    "num_train_epochs": 10,
    "temperature": 21
  }
}