{
  "best_metric": 0.8336594911937377,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-13/checkpoint-672",
  "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": 2.201279878616333,
      "learning_rate": 0.00022445268667640978,
      "loss": 0.5435,
      "step": 96
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7700587084148728,
      "eval_f1": 0.7979363714531386,
      "eval_loss": 0.4716385006904602,
      "eval_precision": 0.7116564417177914,
      "eval_recall": 0.9080234833659491,
      "eval_runtime": 24.8269,
      "eval_samples_per_second": 41.165,
      "eval_steps_per_second": 1.289,
      "step": 96
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.4069817066192627,
      "learning_rate": 0.00018704390556367485,
      "loss": 0.4642,
      "step": 192
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7954990215264188,
      "eval_f1": 0.819047619047619,
      "eval_loss": 0.4382912516593933,
      "eval_precision": 0.734472049689441,
      "eval_recall": 0.9256360078277887,
      "eval_runtime": 25.9254,
      "eval_samples_per_second": 39.421,
      "eval_steps_per_second": 1.234,
      "step": 192
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.9317843914031982,
      "learning_rate": 0.00014963512445093986,
      "loss": 0.4359,
      "step": 288
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8170254403131115,
      "eval_f1": 0.8313796212804329,
      "eval_loss": 0.42131203413009644,
      "eval_precision": 0.7709030100334449,
      "eval_recall": 0.9021526418786693,
      "eval_runtime": 25.4698,
      "eval_samples_per_second": 40.126,
      "eval_steps_per_second": 1.256,
      "step": 288
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.5647993087768555,
      "learning_rate": 0.00011222634333820489,
      "loss": 0.4128,
      "step": 384
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8268101761252447,
      "eval_f1": 0.8338028169014083,
      "eval_loss": 0.4021822512149811,
      "eval_precision": 0.8014440433212996,
      "eval_recall": 0.8688845401174168,
      "eval_runtime": 26.136,
      "eval_samples_per_second": 39.103,
      "eval_steps_per_second": 1.224,
      "step": 384
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.2295299768447876,
      "learning_rate": 7.481756222546993e-05,
      "loss": 0.4001,
      "step": 480
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8287671232876712,
      "eval_f1": 0.84472049689441,
      "eval_loss": 0.40523552894592285,
      "eval_precision": 0.7727272727272727,
      "eval_recall": 0.9315068493150684,
      "eval_runtime": 25.207,
      "eval_samples_per_second": 40.544,
      "eval_steps_per_second": 1.269,
      "step": 480
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.747401475906372,
      "learning_rate": 3.7408781112734964e-05,
      "loss": 0.3945,
      "step": 576
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8326810176125244,
      "eval_f1": 0.8471849865951743,
      "eval_loss": 0.4046291410923004,
      "eval_precision": 0.7796052631578947,
      "eval_recall": 0.9275929549902152,
      "eval_runtime": 25.0049,
      "eval_samples_per_second": 40.872,
      "eval_steps_per_second": 1.28,
      "step": 576
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.8298261165618896,
      "learning_rate": 0.0,
      "loss": 0.3889,
      "step": 672
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8336594911937377,
      "eval_f1": 0.8431734317343174,
      "eval_loss": 0.39929264783859253,
      "eval_precision": 0.7975567190226877,
      "eval_recall": 0.8943248532289628,
      "eval_runtime": 25.0985,
      "eval_samples_per_second": 40.72,
      "eval_steps_per_second": 1.275,
      "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": 0.8348768037192943,
    "learning_rate": 0.0002618614677891448,
    "num_train_epochs": 7,
    "temperature": 10
  }
}