List your company

Spurious Correlation Removal on SAEBench Spurious Correlation Removal Pythia-70M activations

2.1Bias (Professor vs Nurse)Ensembling (NB)
1.923.1354.355.565May 21, 2025

Evaluation results

MethodLinks
Ensembling (NB)Ensemble size=8, Approach=Naive Bagging2025.052.10.41.40.3
Expanded SAE2025.054.70.63.72.1
Ensembling (Boosting)Ensemble size=8, Approach=Boosting2025.056.61.34.52.9