{
  "best_metric": 0.8268101761252447,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-3/checkpoint-288",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 672,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.2690824270248413,
      "learning_rate": 0.00018636956997148543,
      "loss": 0.4818,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.786692759295499,
      "eval_f1": 0.8039568345323741,
      "eval_loss": 0.4252791106700897,
      "eval_precision": 0.7437603993344426,
      "eval_recall": 0.8747553816046967,
      "eval_runtime": 31.9977,
      "eval_samples_per_second": 31.94,
      "eval_steps_per_second": 1.0,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9796446561813354,
      "learning_rate": 0.00015530797497623786,
      "loss": 0.4169,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7935420743639922,
      "eval_f1": 0.8117752007136486,
      "eval_loss": 0.4000055491924286,
      "eval_precision": 0.7459016393442623,
      "eval_recall": 0.8904109589041096,
      "eval_runtime": 30.9158,
      "eval_samples_per_second": 33.058,
      "eval_steps_per_second": 1.035,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.9656918048858643,
      "learning_rate": 0.00012424637998099028,
      "loss": 0.3978,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8268101761252447,
      "eval_f1": 0.8398190045248868,
      "eval_loss": 0.38850346207618713,
      "eval_precision": 0.7811447811447811,
      "eval_recall": 0.9080234833659491,
      "eval_runtime": 30.6883,
      "eval_samples_per_second": 33.303,
      "eval_steps_per_second": 1.043,
      "step": 288
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.4708027839660645,
      "learning_rate": 9.318478498574272e-05,
      "loss": 0.3849,
      "step": 384
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8101761252446184,
      "eval_f1": 0.8098039215686275,
      "eval_loss": 0.3813592493534088,
      "eval_precision": 0.8113948919449901,
      "eval_recall": 0.8082191780821918,
      "eval_runtime": 31.54,
      "eval_samples_per_second": 32.403,
      "eval_steps_per_second": 1.015,
      "step": 384
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.0931408405303955,
      "learning_rate": 6.212318999049514e-05,
      "loss": 0.3774,
      "step": 480
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8238747553816047,
      "eval_f1": 0.8345588235294117,
      "eval_loss": 0.3782808482646942,
      "eval_precision": 0.7868284228769498,
      "eval_recall": 0.8884540117416829,
      "eval_runtime": 31.4899,
      "eval_samples_per_second": 32.455,
      "eval_steps_per_second": 1.016,
      "step": 480
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.7833588123321533,
      "learning_rate": 3.106159499524757e-05,
      "loss": 0.3744,
      "step": 576
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8209393346379648,
      "eval_f1": 0.8325709057639525,
      "eval_loss": 0.37904295325279236,
      "eval_precision": 0.781786941580756,
      "eval_recall": 0.8904109589041096,
      "eval_runtime": 31.8399,
      "eval_samples_per_second": 32.098,
      "eval_steps_per_second": 1.005,
      "step": 576
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.6321544647216797,
      "learning_rate": 0.0,
      "loss": 0.3712,
      "step": 672
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8228962818003914,
      "eval_f1": 0.829406220546654,
      "eval_loss": 0.3763904869556427,
      "eval_precision": 0.8,
      "eval_recall": 0.8610567514677103,
      "eval_runtime": 32.2271,
      "eval_samples_per_second": 31.712,
      "eval_steps_per_second": 0.993,
      "step": 672
    }
  ],
  "logging_steps": 500,
  "max_steps": 672,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 1649866380960.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.6895915406036343,
    "learning_rate": 0.000217431164966733,
    "num_train_epochs": 7,
    "temperature": 20
  }
}