CV-OPD：对比式视觉在线策略蒸馏
CV-OPD: Contrastive Vision On-policy Distillation
AI总结：
针对在线策略蒸馏（OPD）导致视觉语言模型（VLM）依赖捷径的问题，提出CV-OPD方法，通过对比散度损失与差异感知门缓解捷径，提升VLM视觉推理能力。
摘要：
在线策略蒸馏（OPD）在视觉语言模型（VLM）的后训练阶段展现出通用性与便捷性，能从教师模型为学生模型提供密集监督信号。然而我们观察到一个异常现象：当相关视觉区域被掩码或替换为噪声时，经蒸馏得到的VLM仍能准确回答问题。基于该观察，我们的定量与定性实验表明，以答案作为提示的OPD未能教会学生模型真正去寻找并依赖视觉证据。我们的理论分析显示，该问题源于蒸馏过程中缺乏视觉约束所导致的捷径。为解决此问题，我们提出对比式视觉在线策略蒸馏（CV-OPD），该方法在相同学生模型前缀的条件下，对比由正样本和负样本裁剪得到的教师模型分布，并通过跨 token 级和样本级的差异门控，用基于间隔的对比约束来增强优化。具体而言，CV-OPD 首先引入与答案无关的负教师模型，并设计对比散度损失（CDL），通过对比正、负教师模型来约束学生模型的特征分布，纠正学生模型利用视觉区域的方式；CV-OPD 还引入差异感知门（DAG），利用 token 级和样本级的差异选择与视觉相关的 token 进行优化，引导学生模型的注意力转向视觉内容。在广泛的细粒度视觉基准上开展的大量实验表明，CV-OPD 大幅缓解了捷径问题，并持续提升现有VLM的视觉推理能力。
英文摘要：
On-policy distillation (OPD) reveals versatility and convenience during VLM post-training, providing dense supervision for student from teacher. However, we observed an anomaly that the distilled VLM could still answer the questions accurately even when the related visual regions were masked or replaced with noise. Building on this observation, our quantitative and qualitative experiments demonstrate that OPD, which uses the answer as a hint, has failed to teach students to genuinely seek out and rely on visual evidence. Our theoretical analysis suggests that this failure arises from a shortcut enabled by the lack of visual constraints during distillation. To address the issue, we introduce Contrastive Vision On-policy Distillation (CV-OPD), which compares teacher distributions induced by positive and negative crops while conditioning on the same student prefix and augments optimization with a margin-based contrastive constraint, which is modulated with discrepancy gating across token and sample level. Specifically, CV-OPD first introduces a negative teacher unrelated to the answer and designs a Contrastive Divergence Loss (CDL) that constrains the feature distribution of student by contrasting the positive and negative teachers, correcting how the student exploits visual regions. The CV-OPD introduces a Discrepancy Aware Gate (DAG) that uses token-level and sample-level discrepancies to select visually relevant tokens for optimization, directing the student's attention toward visual content. Extensive experiments across a broad range of fine-grained visual benchmarks demonstrate that CV-OPD substantially alleviates shortcut and consistently improves the visual reasoning capabilities of existing VLMs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
