run-27/checkpoint-156/trainer_state.json

{
  "best_metric": 0.7778864970645792,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-27/checkpoint-156",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 156,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 3.795090675354004,
      "learning_rate": 0.0004299539034695828,
      "loss": 0.5539,
      "step": 52
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7397260273972602,
      "eval_f1": 0.7779632721202002,
      "eval_loss": 0.49282991886138916,
      "eval_precision": 0.6783114992721979,
      "eval_recall": 0.9119373776908023,
      "eval_runtime": 28.2474,
      "eval_samples_per_second": 36.18,
      "eval_steps_per_second": 1.133,
      "step": 52
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.3241050243377686,
      "learning_rate": 0.0003582949195579857,
      "loss": 0.4958,
      "step": 104
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.776908023483366,
      "eval_f1": 0.8074324324324325,
      "eval_loss": 0.4643081724643707,
      "eval_precision": 0.7102526002971769,
      "eval_recall": 0.9354207436399217,
      "eval_runtime": 28.4313,
      "eval_samples_per_second": 35.946,
      "eval_steps_per_second": 1.126,
      "step": 104
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.4190268516540527,
      "learning_rate": 0.00028663593564638853,
      "loss": 0.4714,
      "step": 156
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7778864970645792,
      "eval_f1": 0.8090832632464255,
      "eval_loss": 0.4609759449958801,
      "eval_precision": 0.7094395280235988,
      "eval_recall": 0.9412915851272016,
      "eval_runtime": 28.809,
      "eval_samples_per_second": 35.475,
      "eval_steps_per_second": 1.111,
      "step": 156
    }
  ],
  "logging_steps": 500,
  "max_steps": 364,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 707085591840.0,
  "train_batch_size": 59,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.845201757467621,
    "learning_rate": 0.00050161288738118,
    "num_train_epochs": 7,
    "per_device_train_batch_size": 59,
    "temperature": 21
  }
}