run-48/checkpoint-642/trainer_state.json

{
  "best_metric": 0.760398359695372,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-48/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": 6.522511005401611,
      "learning_rate": 0.00018041478341026607,
      "loss": 0.5413,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7123608670181605,
      "eval_f1": 0.47821466524973427,
      "eval_loss": 0.5002033710479736,
      "eval_mcc": 0.3040293618143382,
      "eval_precision": 0.6048387096774194,
      "eval_recall": 0.3954305799648506,
      "eval_runtime": 3.1447,
      "eval_samples_per_second": 542.817,
      "eval_steps_per_second": 17.172,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.938706398010254,
      "learning_rate": 0.00015786293548398283,
      "loss": 0.4858,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7428236672524897,
      "eval_f1": 0.4959816303099885,
      "eval_loss": 0.47951367497444153,
      "eval_mcc": 0.3755947104077684,
      "eval_precision": 0.7152317880794702,
      "eval_recall": 0.37961335676625657,
      "eval_runtime": 3.8209,
      "eval_samples_per_second": 446.753,
      "eval_steps_per_second": 14.133,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8805999755859375,
      "learning_rate": 0.00013531108755769955,
      "loss": 0.4695,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.760398359695372,
      "eval_f1": 0.5530054644808744,
      "eval_loss": 0.4722769260406494,
      "eval_mcc": 0.4255669980344537,
      "eval_precision": 0.7312138728323699,
      "eval_recall": 0.4446397188049209,
      "eval_runtime": 3.1494,
      "eval_samples_per_second": 542.006,
      "eval_steps_per_second": 17.146,
      "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": 0.8927351537481499,
    "learning_rate": 0.00020296663133654935,
    "num_train_epochs": 9,
    "temperature": 2
  }
}