Characterizing Mechanistic Uniqueness and Identifiability Through Circuit Analysis
This is the accompanying code to the paper "Characterizing Mechanistic Uniqueness and Identifiability Through Circuit Analysis," to appear at the 2026 ICLR Workshop on Representational Alignment.
Our circuit enumeration procedure was adapted from the paper, "Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable?" (Meloux et. al, 2025). The paper is available here and the corresponding codebase can be found on Github.