Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

LoMC: Localized Multidirectional Correction for Refusal Suppression in Routed Foundation Models

About

We study controlled post-training refusal suppression in routed MoE and hybrid-MoE foundation models, aiming to increase non-refusal target-response behavior while preserving general capability under a compact intervention footprint. Existing broad direction-based edits can perturb general-purpose computation, whereas support-only expert edits often lack sufficient capacity to correct heterogeneous refusal representations. To address this limitation, we introduce Localized Multidirectional Correction (LoMC), a support-gated intervention framework that follows a support-then-correction execution order: it first identifies a compact edit support, then aggregates prototype correction directions into layer-wise correction directions, and finally applies rank-one layer-wise correction only within the selected support. By using the edit support as a structural gating constraint, LoMC increases correction capacity without expanding the intervention scope. Experiments on text-only and multimodal safety benchmarks across four routed backbones show that LoMC substantially improves non-refusal target-response behavior while maintaining general capability under a compact intervention footprint.

Yan Hong, Kedong Xiu, Wei Li, Jun Lan, Huijia Zhu, Shuheng Zhou, Zhongcai Lyu, Weiqiang Wang, Jianfu Zhang• 2026

Related benchmarks

TaskDatasetResultRank
Refusal suppressionStrongREJECT
TCR100
20
Refusal suppressionJailbreakV-28K
TCR99.12
20
Refusal suppressionVLSBench
TCR100
20
Refusal suppressionAdvBench
Targeted Content Rate (TCR)89.17
20
Showing 4 of 4 rows

Other info

Follow for update