run-43/checkpoint-1935/trainer_state.json

{
  "best_metric": 0.8989918573090345,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-43/checkpoint-1935",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1935,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.1614691019058228,
      "learning_rate": 0.00013740809545058623,
      "loss": 0.2838,
      "step": 645
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8960837533927879,
      "eval_f1": 0.1878787878787879,
      "eval_loss": 0.26379355788230896,
      "eval_mcc": 0.24425925679105032,
      "eval_precision": 0.6813186813186813,
      "eval_recall": 0.10896309314586995,
      "eval_runtime": 9.9542,
      "eval_samples_per_second": 518.172,
      "eval_steps_per_second": 16.274,
      "step": 645
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.5016854405403137,
      "learning_rate": 0.00011450674620882187,
      "loss": 0.2634,
      "step": 1290
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8982163629313688,
      "eval_f1": 0.18856259659969088,
      "eval_loss": 0.26019662618637085,
      "eval_mcc": 0.2656935582466473,
      "eval_precision": 0.782051282051282,
      "eval_recall": 0.10720562390158173,
      "eval_runtime": 9.9403,
      "eval_samples_per_second": 518.897,
      "eval_steps_per_second": 16.297,
      "step": 1290
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7783706784248352,
      "learning_rate": 9.16053969670575e-05,
      "loss": 0.2592,
      "step": 1935
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8989918573090345,
      "eval_f1": 0.187207488299532,
      "eval_loss": 0.2548668682575226,
      "eval_mcc": 0.2745959396731611,
      "eval_precision": 0.8333333333333334,
      "eval_recall": 0.1054481546572935,
      "eval_runtime": 9.9225,
      "eval_samples_per_second": 519.829,
      "eval_steps_per_second": 16.327,
      "step": 1935
    }
  ],
  "logging_steps": 500,
  "max_steps": 4515,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 500,
  "total_flos": 4760827597800.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.7868511609350549,
    "learning_rate": 0.00016030944469235062,
    "num_train_epochs": 7,
    "temperature": 12
  }
}