Is This the Subspace You Are Looking for? An Interpretability Illusion for Subspace Activation Patching

Open Webpage

Aleksandar Makelov, Georg Lange, Atticus Geiger, Neel Nanda

Published: 01 Jan 2024, Last Modified: 15 May 2025ICLR 2024EveryoneRevisionsBibTeXCC BY-SA 4.0