arxiv6.0 / 10

Understanding MoE Decisions Beyond Gradients

Zhen Xu, Zitao Zhang, Kevin Zhao, Xuefeng Liu

Abstract

Understanding a language model's decision requires knowing how its internal computations affect its preference between answers. Gradient-based methods estimate this influence by predicting the effect of changing those computations. In mixture-of-experts (MoE) models, replacing one expert can change which experts later layers select. Those switches can alter the answer scores in ways that derivatives do not capture. We introduce a predictor that uses the model's own expert outputs and selection rules to estimate these extra effects, then adds them to a second-order prediction. Across questions from three benchmarks, our method reduces the error in predicting answer-score changes by 60\% on Qwen3.6 and 47\% on Gemma4, relative to a multistep baseline using gradients and second derivatives.

Research area

causal interventioninterpretabilitymechanistic interpretability
Published
19 Nov 2026
Source
arxiv
Org
University of Chicago
View paper
Sign in to read and join the discussion.