run-19/checkpoint-177/trainer_state.json

{
  "best_metric": 0.7583170254403131,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-19/checkpoint-118",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 177,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.9501977562904358,
      "learning_rate": 0.00014305210915070167,
      "loss": 0.4807,
      "step": 59
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7221135029354208,
      "eval_f1": 0.7171314741035855,
      "eval_loss": 0.42745766043663025,
      "eval_precision": 0.7302231237322515,
      "eval_recall": 0.7045009784735812,
      "eval_runtime": 28.9792,
      "eval_samples_per_second": 35.267,
      "eval_steps_per_second": 1.104,
      "step": 59
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8899472951889038,
      "learning_rate": 9.536807276713445e-05,
      "loss": 0.4215,
      "step": 118
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7583170254403131,
      "eval_f1": 0.7576054955839059,
      "eval_loss": 0.39966535568237305,
      "eval_precision": 0.7598425196850394,
      "eval_recall": 0.7553816046966731,
      "eval_runtime": 27.7652,
      "eval_samples_per_second": 36.809,
      "eval_steps_per_second": 1.153,
      "step": 118
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.816827654838562,
      "learning_rate": 4.7684036383567226e-05,
      "loss": 0.406,
      "step": 177
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7544031311154599,
      "eval_f1": 0.7420349434737924,
      "eval_loss": 0.39147794246673584,
      "eval_precision": 0.7813852813852814,
      "eval_recall": 0.7064579256360078,
      "eval_runtime": 28.1332,
      "eval_samples_per_second": 36.327,
      "eval_steps_per_second": 1.137,
      "step": 177
    }
  ],
  "logging_steps": 500,
  "max_steps": 236,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 707085591840.0,
  "train_batch_size": 52,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.6660099026576007,
    "learning_rate": 0.0001907361455342689,
    "num_train_epochs": 4,
    "per_device_train_batch_size": 52,
    "temperature": 20
  }
}