Self-Routing RAG: Binding Selective Retrieval with Knowledge Verbalization
Di Wu, Jia-Chen Gu, Kai-Wei Chang, Nanyun Peng · 2025
Read Self-Routing RAG: Binding Selective Retrieval with Knowledge Verbalization by Di Wu, Jia-Chen Gu, Kai-Wei Chang, Nanyun Peng (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-01018.txt.
Description
Di Wu, Jia-Chen Gu, Kai-Wei Chang, and Nanyun Peng present Self-Routing RAG, a retrieval system that decides for itself where an answer should come from, in this 2025 arXiv paper. Older systems face a crude choice: fetch from one outside source, or trust the model blindly. Their design treats the model's own knowledge as one source among many.
In a single pass, the system picks a source, may speak its stored knowledge aloud, and then answers from what it picked. A small policy store helps it judge its own uncertainty. Across four benchmarks and three mid-size models, it beat a strong rival while fetching far less often. It also held its accuracy without per-dataset tuning.
Catalog record
| CREATOR | Di Wu, Jia-Chen Gu, Kai-Wei Chang, Nanyun Peng |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-01018