{
  "best_metric": 0.6666666666666666,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-0/checkpoint-160",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 480,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8234807848930359,
      "learning_rate": 6.749457389390413e-05,
      "loss": 0.0587,
      "step": 160
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6666666666666666,
      "eval_f1": 0.0,
      "eval_loss": 0.0437474325299263,
      "eval_mcc": 0.0,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 1.868,
      "eval_samples_per_second": 682.566,
      "eval_steps_per_second": 21.414,
      "step": 160
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.42042386531829834,
      "learning_rate": 3.374728694695207e-05,
      "loss": 0.0451,
      "step": 320
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6650980392156862,
      "eval_f1": 0.04044943820224719,
      "eval_loss": 0.042260926216840744,
      "eval_mcc": 0.031242528987421626,
      "eval_precision": 0.45,
      "eval_recall": 0.021176470588235293,
      "eval_runtime": 1.9237,
      "eval_samples_per_second": 662.79,
      "eval_steps_per_second": 20.793,
      "step": 320
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.15567025542259216,
      "learning_rate": 0.0,
      "loss": 0.0436,
      "step": 480
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6658823529411765,
      "eval_f1": 0.05752212389380531,
      "eval_loss": 0.04147454351186752,
      "eval_mcc": 0.046225016352102424,
      "eval_precision": 0.48148148148148145,
      "eval_recall": 0.03058823529411765,
      "eval_runtime": 1.9554,
      "eval_samples_per_second": 652.025,
      "eval_steps_per_second": 20.456,
      "step": 480
    }
  ],
  "logging_steps": 500,
  "max_steps": 480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 875266313760.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.04540173695128236,
    "learning_rate": 0.00010124186084085621,
    "num_train_epochs": 3,
    "temperature": 16
  }
}