Training-free IN-Retriever uses embedding attention to navigate multi-hop RAG queries
_reachsumit · x · 2026-11-18
An EMNLP 2026 Findings paper proposes IN-Retriever, a training-free multi-hop retrieval method for RAG. Instead of generating intermediate reasoning, it extracts bridging clues from the embedding model's internal attention, strips revealed clues from the query vector, and reranks candidates to steer retrieval toward hidden next-hop evidence. It outperforms sparse and dense retrieval baselines.
More from Research
- UCLA hosts California AI Amplitudes school applying ML to quantum scattering amplitudes — KyleCranmer · 2026-11-18
- Looped layers don't help multimodal embeddings unless recurrence is pretrained (+4.23 points) — _reachsumit · 2026-11-18
- Ant Group's Repo-Premium shows BM25 matches dense retrieval in fair code-search tests — _reachsumit · 2026-11-18
- Kuaishou proposes GBPO, fixing GRPO's dead gradients and unbounded updates for generative recommenders — _reachsumit · 2026-11-18
- GainReason rewards each reasoning sentence by its predictive gain for generative recommendation — _reachsumit · 2026-11-18
- Tencent's TokenFormer V2 matches top unified recommenders with 5.1× fewer FLOPs — _reachsumit · 2026-11-18