#artificial intelligence
Sep 2025
Steering Multimodal Large Language Models Decoding for Context-Aware Safety
Safety-aware Contrastive Decoding (SafeCoDe) is introduced, a lightweight and model-agnostic decoding framework that dynamically adjusts token generation based on multimodal context that consistently improves context-sensitive refusal behaviors while preserving model helpfulness.
Zheyuan Liu, Zhangchen Xu, Guangyao Dou et al.
· arXiv.org · 6 citations