run-41/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7428236672524897,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-41/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": 7.4478278160095215,
      "learning_rate": 0.00011476430024641456,
      "loss": 0.5267,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7094317516110135,
      "eval_f1": 0.487603305785124,
      "eval_loss": 0.4861217737197876,
      "eval_mcc": 0.3024497025688451,
      "eval_precision": 0.5914786967418546,
      "eval_recall": 0.4147627416520211,
      "eval_runtime": 3.1538,
      "eval_samples_per_second": 541.244,
      "eval_steps_per_second": 17.122,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.338618755340576,
      "learning_rate": 5.738215012320728e-05,
      "loss": 0.4743,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7381370826010545,
      "eval_f1": 0.5299684542586751,
      "eval_loss": 0.4651449918746948,
      "eval_mcc": 0.37172090324876994,
      "eval_precision": 0.6596858638743456,
      "eval_recall": 0.4428822495606327,
      "eval_runtime": 3.7183,
      "eval_samples_per_second": 459.08,
      "eval_steps_per_second": 14.523,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6570167541503906,
      "learning_rate": 0.0,
      "loss": 0.4594,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7428236672524897,
      "eval_f1": 0.5334750265674814,
      "eval_loss": 0.46071532368659973,
      "eval_mcc": 0.3822943460437718,
      "eval_precision": 0.6747311827956989,
      "eval_recall": 0.44112478031634444,
      "eval_runtime": 3.1598,
      "eval_samples_per_second": 540.217,
      "eval_steps_per_second": 17.089,
      "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": 0.8513780187435334,
    "learning_rate": 0.00017214645036962186,
    "num_train_epochs": 3,
    "temperature": 16
  }
}