Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

arXiv:2403.19647v1 Announce Type: new
Abstract: We introduce methods for discovering and applying sparse feature circuits. These are causally implicated subnetworks of human-interpretable features for explaining language model behaviors. Circuits identified in prior work consist of polysemantic and difficult-to-interpret units like attention heads or neurons, rendering them unsuitable for many downstream applications. In contrast, sparse feature circuits enable detailed understanding of unanticipated mechanisms. Because they are based on fine-grained units, sparse feature circuits are useful for downstream tasks: We introduce SHIFT, …

abstract arxiv attention causal circuits cs.ai cs.cl cs.lg editing feature features graphs human language language model language models neurons prior rendering them type units work