arxivcs.LG2026-06-30
Representation as a Bottleneck for Mechanistic Interpretability: The Manifestation Unit Protocol
Hussein Chouman, Wataru Sasaki, Tomokazu Matsui, Hirohiko Suwa, Keiichi Yasumoto
Mechanistic interpretability has produced a rich inventory of component-level analyses that characterise what neural-network components encode and how they interact. Their outputs, however, are not easily reusable: selectivity tables, circuit diagrams, and feature lists remain lo…