Training-free Seek framework beats BM25 by 82% on BRIGHT via self-evaluative iteration
_reachsumit · x · 2026-09-25
Seek, a training-free retrieval framework accepted at CIKM 2026, addresses the single-pass limitation of LLM retrievers and rerankers, where missed documents are permanently lost.
Each round, an LLM generates pseudo-passages conditioned on accumulated relevance feedback, a retriever surfaces fresh candidates, and an assessor assigns graded relevance to guide subsequent rounds.
- Matches trained rerankers on TREC Deep Learning while consistently improving Recall@100 over single-pass BM25
- On reasoning-intensive BRIGHT, Seek with Qwen2.5-7B gains 82% over BM25, beating all trained baselines
- With GPT-4.1 it reaches 37.4 average nDCG@10, exceeding the strongest baseline by 37%
More from Research
- PosteriorBench accepted to NeurIPS: point reconstruction accuracy misleads on inverse solvers — AnimaAnandkumar · 2026-09-25
- Dietterich: pretraining's next-token task already forces models to abstract and generalize — tdietterich · 2026-09-25
- Neural Spectral Capacity: a training-free architecture score that beats #Params and prunes LLaMA-7B 5900x faster — CityU-HongKong · 2026-09-25
- HRI 2027 Adds Archival Industry White Paper Track for Real-World Robot Deployments — petitegeek · 2026-09-25
- Study: Reasoning hurts 15.7% of multimodal embeddings; training-free SURE router fixes it — _reachsumit · 2026-09-25
- Google scales learned cross-task relationships in YouTube's production recommender — _reachsumit · 2026-09-25