{
  "best_metric": 0.8268101761252447,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-11/checkpoint-240",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 336,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.152926206588745,
      "learning_rate": 0.0007816128679887146,
      "loss": 0.591,
      "step": 48
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7524461839530333,
      "eval_f1": 0.7944760357432982,
      "eval_loss": 0.4880000948905945,
      "eval_precision": 0.6791666666666667,
      "eval_recall": 0.9569471624266145,
      "eval_runtime": 29.551,
      "eval_samples_per_second": 34.584,
      "eval_steps_per_second": 0.541,
      "step": 48
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.895570278167725,
      "learning_rate": 0.0006513440566572622,
      "loss": 0.4603,
      "step": 96
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8199608610567515,
      "eval_f1": 0.8286778398510243,
      "eval_loss": 0.4088345766067505,
      "eval_precision": 0.7904085257548845,
      "eval_recall": 0.8708414872798435,
      "eval_runtime": 30.3763,
      "eval_samples_per_second": 33.645,
      "eval_steps_per_second": 0.527,
      "step": 96
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.8208940029144287,
      "learning_rate": 0.0005210752453258098,
      "loss": 0.4269,
      "step": 144
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8140900195694716,
      "eval_f1": 0.8327464788732395,
      "eval_loss": 0.45197436213493347,
      "eval_precision": 0.7568,
      "eval_recall": 0.9256360078277887,
      "eval_runtime": 29.5919,
      "eval_samples_per_second": 34.536,
      "eval_steps_per_second": 0.541,
      "step": 144
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.270200729370117,
      "learning_rate": 0.0003908064339943573,
      "loss": 0.3888,
      "step": 192
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8189823874755382,
      "eval_f1": 0.8262910798122065,
      "eval_loss": 0.40383195877075195,
      "eval_precision": 0.7942238267148014,
      "eval_recall": 0.8610567514677103,
      "eval_runtime": 30.1222,
      "eval_samples_per_second": 33.928,
      "eval_steps_per_second": 0.531,
      "step": 192
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.684295415878296,
      "learning_rate": 0.0002605376226629049,
      "loss": 0.3669,
      "step": 240
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8268101761252447,
      "eval_f1": 0.8409703504043126,
      "eval_loss": 0.41744667291641235,
      "eval_precision": 0.7774086378737541,
      "eval_recall": 0.9158512720156555,
      "eval_runtime": 29.2793,
      "eval_samples_per_second": 34.905,
      "eval_steps_per_second": 0.546,
      "step": 240
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.6306262016296387,
      "learning_rate": 0.00013026881133145244,
      "loss": 0.3718,
      "step": 288
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8189823874755382,
      "eval_f1": 0.8378615249780893,
      "eval_loss": 0.40002480149269104,
      "eval_precision": 0.7587301587301587,
      "eval_recall": 0.9354207436399217,
      "eval_runtime": 29.4445,
      "eval_samples_per_second": 34.709,
      "eval_steps_per_second": 0.543,
      "step": 288
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.835477828979492,
      "learning_rate": 0.0,
      "loss": 0.348,
      "step": 336
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8228962818003914,
      "eval_f1": 0.8344007319304666,
      "eval_loss": 0.40344515442848206,
      "eval_precision": 0.7835051546391752,
      "eval_recall": 0.8923679060665362,
      "eval_runtime": 29.1711,
      "eval_samples_per_second": 35.035,
      "eval_steps_per_second": 0.548,
      "step": 336
    }
  ],
  "logging_steps": 500,
  "max_steps": 336,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 1649866380960.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9971844266637861,
    "learning_rate": 0.0009118816793201671,
    "num_train_epochs": 7,
    "temperature": 16
  }
}