发表机构
Beijing University of Posts and Telecommunications; ByteDance; USTC; Beijing Key Laboratory of Network System and Network Culture; Key Laboratory of Interactive Technology and Experience System, Ministry of Culture and Tourism; Zhongguancun Academy（北京邮电大学; 字节跳动; 中国科学技术大学; 北京网络系统与网络文化重点实验室; 文化和旅游部互动技术与体验系统重点实验室; 中关村科学城）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究多模态推理的在线策略蒸馏问题，提出H-OPD框架，通过验证异构教师互补性，以令牌级教师仲裁取代任务或样本级路由，结合视觉与文本教师，在多基准测试中性能优越。
AI 中文摘要
在线策略蒸馏（OPD）是一种有效的训练后范式。现有多模态推理的OPD方法依赖静态教师路由，忽略不同解码步骤主导模态不同。为此提出H-OPD，通过验证异构教师互补性，用令牌级教师仲裁取代任务或样本级路由，采用视觉到语言描述转移，在各令牌处动态组合教师，多基准测试验证了其优越性能。
英文摘要
On-policy distillation (OPD) has recently emerged as an effective post-training paradigm by providing supervision on student-generated trajectories. However, existing OPD methods for multimodal reasoning usually rely on a static teacher routing, assigning each sample to a single teacher based on modality or task type. This ignores that visual grounding and abstract reasoning may dominate different decoding steps, making a single teacher insufficient for the full trajectory. To this end, H-OPD is proposed as a confidence-aware heterogeneous multi-teacher OPD framework for multimodal reasoning. By verifying the complementarity of heterogeneous teachers in the same reasoning process, H-OPD replaces task or sample level teacher routing with token-level teacher arbitration along the shared student trajectory. H-OPD employs vision-to-language description transfer to enable text-only teachers to access key visual semantics, and uses a confidence-aware arbitration mechanism to dynamically combine vision-language teacher and text-only teachers at each token. Extensive evaluations over 11 widely-used reasoning benchmarks showcase the superior performance of our method.
CommentsEMNLP2026
