提出CrossQ量化方法，让检索模型在压缩后仍保持高精度。
CrossQ: Task-Aligned Cross-Token Conditional Quantization for Late Interaction Retrieval
- 根据文档上下文动态调整令牌编码，提升关键信息保真度。
- 在2B/token下，相比最强基线，MRR@10提升0.010。
- 压缩率高达64倍，适合内存受限的实时检索场景。
晚交互检索器如ColBERT虽精度高，但多向量索引体积大。传统压缩侧重令牌重建误差最小化，而排序依赖于稀疏“胜出”令牌的得分保留。我们提出CrossQ，通过在索引时计算轻量文档上下文（不存储）来条件化令牌编码，自适应提升文档内有效令牌保真度。CrossQ采用对齐排名的目标进行训练，以维持候选得分分布并保护难负样本边界。在2B/token下，CrossQ比最强的严格占用匹配量化基线提升MRR@10达+0.010，比最强候选匹配系统参考提升+0.012。在九个BEIR数据集子集上，4B/token下平均nDCG@10比最强候选匹配系统参考提升+0.009。在4B/token时，实现64倍原始令牌存储压缩，含元数据约61倍，保守填充/对齐估算下约58倍。在8B/token时，经轻量微调后，保留约98%全精度ColBERT的MRR@10，显著改善内存受限场景下的性能-占用权衡。
原文摘要 · Abstract (English)
Late-interaction retrievers like ColBERT achieve high quality but suffer from large multi-vector indices. Standard compression minimizes token reconstruction error, while ranking depends critically on preserving scores of sparse "winner" tokens. We introduce CrossQ, which adaptively improves effective token fidelity within documents by conditioning token codes on lightweight document context computed at indexing time (but not stored). CrossQ is trained with ranking-aligned objectives that preserve candidate score distributions and protect hard-negative margins. At 2 B/token, CrossQ improves MRR@10 by +0.010 over the strongest strictly footprint-matched quantization baseline and by +0.012 over the strongest candidate-matched system reference. On a nine-dataset BEIR subset, CrossQ improves average nDCG@10 by +0.009 at 4 B/token over the strongest candidate-matched system reference. At 4 B/token, CrossQ achieves 64x raw token-storage reduction, approximately 61x including metadata and approximately 58x under conservative padding/alignment accounting. At 8 B/token, CrossQ with light fine-tuning retains approximately 98% of full-precision ColBERT MRR@10, improving the footprint-quality tradeoff for memory-constrained late-interaction retrieval.
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。