Study: RAG retrieval diversification helps only on redundant multi-evidence pools, paper proposes per-query rule
_reachsumit · x · 2026-10-08
Researchers from Université du Québec à Trois-Rivières show that retrieval diversification in RAG helps or hurts depending primarily on candidate-pool redundancy and the number of distinct evidence pieces a query needs. It harms relevance, coverage and answer quality on clean pools, but benefits multi-evidence queries where near-duplicates crowd top-k. They propose a query-adaptive rule that diversifies only when the effective number of distinct documents in top-k falls below the query's evidence requirement, computed from existing embeddings and transferable across datasets and encoders, plus RNG-Score, a geometric reranker whose margin indicates duplicate structure.
More from Research
- Building a computer out of fluids: simulating OpenAI's Navier–Stokes Turing machine — kfountou · 2026-10-08
- D-OPCD distills agent harness gains into diffusion weights, lifting T2I scores 60.52 to 65.09 — Wenxuan Wang · 2026-10-08
- QuadTok: quadtree visual tokenizer saves 10% tokens, hits 2.08 gFID on ImageNet 256 — Yucheng Mao · 2026-10-08
- PhysEvo: recursive self-improvement around a frozen model hits 62% on RoboDojo, 84% on real robot — Wenqing Tian · 2026-10-08
- RL perspective explains on-policy distillation collapse via implicit-reward hacking — Han Cui · 2026-10-08
- SketchSSM cuts linear-attention state traffic 10x, speeds decode up to 7.3x on B300 — sehoonkim418 · 2026-10-08