FLINT cuts late-interaction retrieval index by 89% and latency up to 170×
matei_zaharia · x · 2026-10-24
Matei Zaharia's team introduces FLINT (Fixed-budget Late Interaction), which symmetrically pools queries and documents into a fixed number of vectors via attention-based pooling. Results: up to 89% smaller index than ColBERTv2 (5.3-9.3× reduction), 23-170× faster retrieval (51.3ms vs 8.7s on TREC-ToT 2025), on-par or better ranking on BEIR and long-query benchmarks like BrowseComp+ and CRUMB, validated on BERT and ModernBERT backbones. Paper: arXiv 2610.27273.
More from Research
- Vik Paruchuri answers 'self-topped benchmark is suspect' with open audit — VikParuchuri · 2026-10-25
- Datalab benchmarks 17 open OCR models on OmniParseBench; Chandra OCR 2 tops at 90.1 — VikParuchuri · 2026-10-25
- Datalab releases OmniParseBench and ranks 17 open OCR models — VikParuchuri · 2026-10-25
- Proposal: report AUC only, or add a dev set to tune thresholds per eval — antoine_chaffin · 2026-10-25
- Adaptive Skill Granularity Framework Routes Coding Agent Skills at Runtime, Hits 100% Success — dair_ai · 2026-10-25
- Study: LLMs rewriting their own ARC-AGI-3 harnesss gain just 2.8 points, not significant — omarsar0 · 2026-10-25