研究：MoE 模型合并后的路由漂移不足以诊断路由失效

View PDF HTML (experimental)

Abstract:Model merging efficiently combines specialized large language models (LLMs) without joint retraining, but can substantially alter expert routing in Mixture-of-Experts (MoE) models. Such \emph{routing drift} is often interpreted as routing failure, raising a fundamental question that remains unclear: \emph{does routing drift after MoE merging actually indicate routing failure, and what evidence should justify repair?} We investigate these questions across DeepSeekMoE, OLMoE, and Qwen3-MoE proposing a routing analysis toolkit for controlled counterfactual interventions and token-level analysis. By crossing source and merged router inputs and parameters, we attribute most expert reassignments to input shifts rather than parameter changes at the same layer. However, source-relative routing differences poorly predict next-token likelihood gains from source-route restoration, and different expert selections can produce directionally similar mixture outputs. We therefore operationalize routing failure as \textit{task loss recoverable under a specified routing intervention, with non-routing parameters fixed.} These tests detect recoverable loss under deliberate router corruption, whereas source-route restoration does not establish reliable task benefits in the evaluated merged models. Motivated by these, we propose \emph{Selective Router Repair (SRR)} as a case study, and find that source-specialist token-likelihood advantages do not reliably identify beneficial local corrections. Together, these findings show that \textbf{routing drift alone is insufficient evidence of routing failure}: source-informed corrections must be judged by their task-level intervention effects. The analysis toolkit and SRR code are released.

Submission history

From: Yuanyi Wang [view email]
[v1] Sat, 26 Sep 2026 17:47:55 UTC (1,406 KB)