sbert_cwl_split_60_lambda_0_1/training_meta.json

{
  "model_type": "sbert_cwl_l2",
  "base_model": "bert-base-uncased",
  "train_split": "train_60",
  "lambda_consistency": 0.1,
  "augmentations": [
    "back_translation",
    "aeda"
  ],
  "cwl_selection": "argmax_l2_distance",
  "augmented_dataset": "SurAyush/stsb_augmented_splits",
  "original_dataset": "SurAyush/stsb_splits",
  "pooling": "mean",
  "val_spearman": 0.8718,
  "test_spearman": 0.8306,
  "baseline_bert": 0.4729,
  "sbert_no_reg": 0.8192,
  "delta_vs_bert": 0.3577,
  "delta_vs_sbert": 0.0114,
  "lr": 2e-05,
  "batch_size": 16,
  "early_stop_patience": 2
}