GLInt：晚期交互检索的几何匹配硬负样本挖掘
原标题：GLInt: Geometry-Matched Hard Negatives for Late-Interaction Retrieval
AI 摘要
Hugging Face 博客发布了一篇关于 GLInt 的实地报告，GLInt 是一种用于晚期交互检索的几何匹配硬负样本挖掘方法。作者从多向量检索器挖掘的硬负样本是否优于密集检索器的问题出发，构建了 GLInt 模型，在 BEIR-15 任务上达到 57.43 的平均 nDCG@10，超越了 LateOn 的 57.22，成为 300M 参数以下最强的检索器。报告详细分析了 MaxSim 评分几何特性，发现正样本感知的比率阈值在 MaxSim 下不稳定，因此改用排名相对规则。模型和训练数据已开源。
正文节选
A field report on multi-vector mining, MaxSim geometry, knowledge distillation, and the experiments that did not work. I started with a narrow question: Do hard negatives mined by a multi-vector retriever produce a better late-interaction model than negatives mined by a dense retriever? Pffftt, Obviously! I introduce GLInt, which reaches 57.43 mean nDCG@10 on the 15 BEIR Tasks, starting from the unsupervised LateOn checkpoint. It outperforms the 57.22 result of LateOn, making it the strongest re