A residual learning approach enhances Sparse Autoencoders to capture domain-specific features without retraining, improving interpretability and performance on specialized domains.
Sparse Autoencoders have emerged as powerful tools for interpreting the
internal representations of Large Language Models, yet they often fail to
capture domain-specific features not prevalent in their training corpora. This
paper introduces a residual learning approach that addresses this feature
blindness without requiring complete retraining. We propose training a
secondary SAE specifically to model the reconstruction error of a pretrained
SAE on domain-specific texts, effectively capturing features missed by the
primary model. By summing the outputs of both models during inference, we
demonstrate significant improvements in both LLM cross-entropy and explained
variance metrics across multiple specialized domains. Our experiments show that
this method efficiently incorporates new domain knowledge into existing SAEs
while maintaining their performance on general tasks. This approach enables
researchers to selectively enhance SAE interpretability for specific domains of
interest, opening new possibilities for targeted mechanistic interpretability
of LLMs.