VLM教师能传递什么？视觉文档检索监督信号对比
原标题：What Transfers from a VLM Teacher? Comparing Supervision Signals for Visual Document Retrieval
AI 摘要
该研究比较了在视觉文档检索中如何利用VLM教师模型的不同监督信号。在固定学生模型、数据、优化器和评估的条件下，让教师判断挖掘出的困难负样本（N）和分数蒸馏（S）比丰富正样本的描述对齐（D）和注意力接地（A）更有效，将ViDoRe v2的nDCG@5从基线提升至更高水平。研究还发现训练集中的单正样本标签严重不完整，标注者认为查询前四名挖掘候选中约两个是相关的但未被标注，导致训练将相关页面当作负样本推开。作者发布了代码、教师判断、挖掘池、人工审计和训练适配器。
正文节选
What Transfers from a VLM Teacher? Comparing Supervision Signals for Visual Document Retrieval Abstract Visual document retrievers are trained contrastively: each query is matched to one page labelled relevant — the positive — and pushed away from negatives, pages presumed irrelevant. Recent methods distil a vision–language model (VLM) teacher into the retriever by enriching that positive, transferring the teacher’s attention over it or a description of it. We ask whether the teacher is better s