ColBERT 实战解析:重排再快也救不回检索器漏掉的文档

antoine_chaffin · x · 2026-10-28

A new blog post compares ColBERT's two deployment modes: re-ranking BM25 results vs. end-to-end retrieval.

Key points:

• 2019 BERT rankers beat prior methods by 7 MRR@10 points on MS MARCO (8.8M passages), but score every candidate separately: 10,700ms per query on a V100 for 1,000 candidates.
• ColBERT (Khattab & Zaharia, SIGIR 2020) encodes queries and documents separately with late interaction: MRR@10 of 34.9 vs. 34.7 for BERT-base, at 61ms per query — 170x faster at equal quality.
• The blind spot: re-ranking can only reorder what BM25 hands it. BM25's top-1,000 contains the right passage only 81.4% of the time, so nearly 1 in 5 relevant passages is gone before ColBERT scores anything.
• The fix is to stop relying on BM25 for candidates, leading into the end-to-end trade-offs the article explores.

原文链接 →

「编程与Agent」频道最新

• 微软 Copilot 发布会曝光:Fusion、Projects、按需生成智能体军团 — kimmonismus · 2026-10-29
• 项目经理自述:AI Agent 已自动化其 70% 日常工作 — Theinkling620 · 2026-10-29
• Vercel 工程师分享 video.md:给人和 agent 看的视频生成工作流 — JohnPhamous · 2026-10-29
• Expo 晒 AI 编码打造原生 iPhone 双人应用,发布赶在预约 AppleCare 前 — Dimillian · 2026-10-29
• LangChain:MDA 上线动态调度后 70% 运行来自定时任务 — LangChain · 2026-10-29
• NVIDIA 发布 Metatron:用目标接口让编码智能体驱动人形机器人全身运动 — carlosdponx · 2026-10-29

更多「编程与Agent」频道 AI 资讯 →