ProxySPEX: Inference-Efficient Interpretability via Sparse Feature Interactions in LLMs

Add code
May 23, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: