Revitalizing Black-Box Interpretability: Actionable Interpretability for LLMs via Proxy Models.

Junhao Liu, Haonan Yu, Zhenyu Yan, Xin Zhang

Browse the full ACL paper archive.