Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanisms

arXiv:2403.17806v1 Announce Type: new
Abstract: Many recent language model (LM) interpretability studies have adopted the circuits framework, which aims to find the minimal computational subgraph, or circuit, that explains LM behavior on a given task. Most studies determine which edges belong in a LM's circuit by performing causal interventions on each edge independently, but this scales poorly with model size. Edge attribution patching (EAP), gradient-based approximation to interventions, has emerged as a scalable but imperfect solution to this problem. In …

abstract arxiv behavior beyond causal circuits computational cs.cl cs.lg framework interpretability language language model studies type