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 this paper, we introduce a new method - EAP with integrated gradients (EAP-IG) - that aims to better maintain a core property of circuits: faithfulness. A circuit is faithful if all model edges outside the circuit can be ablated without changing the model's performance on the task; faithfulness is what justifies studying circuits, rather than the full model. Our experiments demonstrate that circuits found using EAP are less faithful than those found using EAP-IG, even though both have high node overlap with circuits found previously using causal interventions. We conclude more generally that when using circuits to compare the mechanisms models use to solve tasks, faithfulness, not overlap, is what should be measured.

翻译：近年来，许多语言模型可解释性研究采用了电路框架，该框架旨在找到解释语言模型在特定任务上行为的最小计算子图（即电路）。大多数研究通过独立对每条边进行因果干预来确定哪些边属于语言模型的电路，但这种方法随模型规模增大而难以扩展。基于梯度的干预近似方法——边归因修补（EAP）——已成为该问题可扩展但不完美的解决方案。本文提出一种新方法——结合积分梯度的EAP（EAP-IG）——旨在更好地维护电路的核心特性：忠实性。若电路外所有模型边被消融后模型在任务上的性能保持不变，则该电路具有忠实性；正是忠实性证明了研究电路（而非完整模型）的合理性。实验表明，尽管EAP发现的电路与先前通过因果干预发现的电路具有高度节点重叠性，但其忠实性低于EAP-IG发现的电路。我们更普遍地得出结论：当使用电路比较模型解决任务的机制时，应衡量的是忠实性而非重叠性。