Think Straight, Stop Smart: Structured Reasoning for Efficient Multi-Hop RAG
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917032299069440 |
|---|---|
| author | Bang, Jihwan Lee, Juntae Yang, Seunghan Choi, Sungha |
| author_facet | Bang, Jihwan Lee, Juntae Yang, Seunghan Choi, Sungha |
| contents | Multi-hop retrieval-augmented generation (RAG) is a promising strategy for complex reasoning, yet existing iterative prompting approaches remain inefficient. They often regenerate predictable token sequences at every step and rely on stochastic stopping, leading to excessive token usage and unstable termination. We propose TSSS (Think Straight, Stop Smart), a structured multi-hop RAG framework designed for efficiency. TSSS introduces (i) a template-based reasoning that caches recurring prefixes and anchors sub-queries to the main question, reducing token generation cost while promoting stable reasoning, and (ii) a retriever-based terminator, which deterministically halts reasoning once additional sub-queries collapse into repetition. This separation of structured reasoning and termination control enables both faster inference and more reliable answers. On HotpotQA, 2WikiMultiHop, and MuSiQue, TSSS achieves state-of-the-art accuracy and competitive efficiency among RAG-CoT approaches, highlighting its effectiveness in efficiency-constrained scenarios such as on-device inference. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_19171 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Think Straight, Stop Smart: Structured Reasoning for Efficient Multi-Hop RAG Bang, Jihwan Lee, Juntae Yang, Seunghan Choi, Sungha Computation and Language Multi-hop retrieval-augmented generation (RAG) is a promising strategy for complex reasoning, yet existing iterative prompting approaches remain inefficient. They often regenerate predictable token sequences at every step and rely on stochastic stopping, leading to excessive token usage and unstable termination. We propose TSSS (Think Straight, Stop Smart), a structured multi-hop RAG framework designed for efficiency. TSSS introduces (i) a template-based reasoning that caches recurring prefixes and anchors sub-queries to the main question, reducing token generation cost while promoting stable reasoning, and (ii) a retriever-based terminator, which deterministically halts reasoning once additional sub-queries collapse into repetition. This separation of structured reasoning and termination control enables both faster inference and more reliable answers. On HotpotQA, 2WikiMultiHop, and MuSiQue, TSSS achieves state-of-the-art accuracy and competitive efficiency among RAG-CoT approaches, highlighting its effectiveness in efficiency-constrained scenarios such as on-device inference. |
| title | Think Straight, Stop Smart: Structured Reasoning for Efficient Multi-Hop RAG |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2510.19171 |