Firehose

Filtered to Papers, tagged “BM25” · clear filters

Browse: People · Companies · Papers · Podcasts · Hacker News · Deep dives

All People Companies Papers Podcasts Hacker News

Browse by tag

continual learning 23 reinforcement learning 9 language models 6 benchmarking 4 natural language processing 4 AI agents 3 vision-language models 3 world models 3

30 JUL 2026 · Paper

BM25 Wins at Scale: A Scaling Study of Retrieval-Augmented Generation Paradigms

This paper compares the performance of different retrieval-augmented generation (RAG) paradigms at varying corpus sizes, finding that BM25 outperforms others at larger scales, but not at smaller ones, and that lexical retrieval is the strongest scalable default.

RAG lexical retrieval BM25 scalability corpus size