run-12/checkpoint-240/trainer_state.json

{
  "best_metric": 0.8258317025440313,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-12/checkpoint-240",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 240,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.1590919494628906,
      "learning_rate": 0.0007554421241694311,
      "loss": 0.5878,
      "step": 48
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7602739726027398,
      "eval_f1": 0.7911338448422848,
      "eval_loss": 0.49036356806755066,
      "eval_precision": 0.7009063444108762,
      "eval_recall": 0.9080234833659491,
      "eval_runtime": 27.1211,
      "eval_samples_per_second": 37.683,
      "eval_steps_per_second": 0.59,
      "step": 48
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.101812362670898,
      "learning_rate": 0.0006475218207166552,
      "loss": 0.4815,
      "step": 96
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7857142857142857,
      "eval_f1": 0.8003646308113036,
      "eval_loss": 0.45267680287361145,
      "eval_precision": 0.7491467576791809,
      "eval_recall": 0.8590998043052838,
      "eval_runtime": 27.6158,
      "eval_samples_per_second": 37.008,
      "eval_steps_per_second": 0.579,
      "step": 96
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.0611319541931152,
      "learning_rate": 0.0005396015172638793,
      "loss": 0.4483,
      "step": 144
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7896281800391389,
      "eval_f1": 0.8151332760103182,
      "eval_loss": 0.47112011909484863,
      "eval_precision": 0.7269938650306749,
      "eval_recall": 0.9275929549902152,
      "eval_runtime": 28.0904,
      "eval_samples_per_second": 36.383,
      "eval_steps_per_second": 0.57,
      "step": 144
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.9773441553115845,
      "learning_rate": 0.0004316812138111035,
      "loss": 0.4206,
      "step": 192
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8131115459882583,
      "eval_f1": 0.8168744007670182,
      "eval_loss": 0.4279371500015259,
      "eval_precision": 0.8007518796992481,
      "eval_recall": 0.8336594911937377,
      "eval_runtime": 27.1892,
      "eval_samples_per_second": 37.588,
      "eval_steps_per_second": 0.588,
      "step": 192
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.5902297496795654,
      "learning_rate": 0.0003237609103583276,
      "loss": 0.3948,
      "step": 240
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8258317025440313,
      "eval_f1": 0.8348794063079776,
      "eval_loss": 0.4194752871990204,
      "eval_precision": 0.7936507936507936,
      "eval_recall": 0.8806262230919765,
      "eval_runtime": 27.2832,
      "eval_samples_per_second": 37.459,
      "eval_steps_per_second": 0.586,
      "step": 240
    }
  ],
  "logging_steps": 500,
  "max_steps": 384,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "total_flos": 1178475986400.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9354133816526636,
    "learning_rate": 0.000863362427622207,
    "num_train_epochs": 8,
    "temperature": 15
  }
}