无需训练的结构锚点剪枝，实现视觉文档检索的高效压缩。
Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval
- 通过自校准框架自动定位可剪枝区域，无需调参。
- 压缩90%视觉标记，仍保持93%-96%检索精度（v1）。
- 适合追求高压缩比且无标注数据的检索系统部署者。
近期视觉语言模型（如ColPali）实现了细粒度视觉文档检索（VDR），但带来了巨大的多向量索引存储开销。现有无训练剪枝方法依赖启发式层选择，或在激进压缩下性能急剧下降，因此先前研究认为有效高压缩剪枝需依赖查询的训练。本文提出结构锚点剪枝（SAP），一种自校准、无训练、查询无关的索引时框架，包含：(i) 分层压缩诊断工具评分保留（SR）；(ii) SR引导的窗口选择，自动定位任意骨干网络的结构剪枝区域，无需模型特定超参数；(iii) 视觉入度中心性评分器，识别该窗口内的锚定图像块。在跨三个架构（18、28、36层）的ViDoRe v1/v2上，SAP在压缩90%视觉标记的情况下，仍保持v1上93–96%的NDCG@5和v2上88–90%；在20×压缩下，分别保持85–90%和76–79%。层级解析的SR分析揭示了对齐-聚合差异：视觉结构在骨干网络中呈现稳定的“结构平台”，而最后几层将其重塑为稀疏、查询对齐的形式，不适宜剪枝。探测预检索基础骨干网络发现，对比微调使这一边界增强三至八倍，解释了为何末层方法失效。
原文摘要 · Abstract (English)
Recent Vision-Language Models (e.g., ColPali) enable fine-grained Visual Document Retrieval (VDR) but incur prohibitive multi-vector index storage overhead. Existing training-free pruning methods either rely on heuristic layer choices or degrade sharply under aggressive compression, leading prior work to argue that effective high-compression pruning requires query-dependent training. We challenge this view with Structural Anchor Pruning (SAP), a self-calibrating, training-free, query-agnostic index-time framework combining (i) Score Retention (SR), a white-box per-layer compression diagnostic; (ii) SR-guided window selection, which automatically locates the structural pruning region of any backbone with no per-model hyperparameters; and (iii) a visual in-degree centrality scorer that identifies anchor patches within that window. On ViDoRe v1/v2 across three architectures spanning 18, 28, and 36 backbone layers, SAP retains 93--96\% of NDCG@5 on v1 and 88--90\% on the harder v2 while pruning 90\% of visual tokens; at 20$\times$ compression it retains 85--90\% and 76--79\% respectively. Our layer-resolved SR analysis reveals an Alignment-Aggregation Divergence: visual structure is preserved as a stable ``Structural Plateau'' within the backbone, while the final layers reshape it into a sparse, query-aligned form unsuitable for pruning. Probing the pre-retrieval base backbones shows that contrastive fine-tuning sharpens this boundary three- to eight-fold, explaining why final-layer methods fail.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。