Graph-Augmented Reasoning: Evolving Step-by-Step Knowledge Graph Retrieval for LLM Reasoning
Wenjie Wu, Yongcheng Jing, Yingjie Wang, Wenbin Hu, Dacheng Tao · 2025
Read Graph-Augmented Reasoning: Evolving Step-by-Step Knowledge Graph Retrieval for LLM Reasoning by Wenjie Wu, Yongcheng Jing, Yingjie Wang, Wenbin Hu, Dacheng Tao (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-2503-01642.txt.
Description
Wenjie Wu, Yongcheng Jing, Yingjie Wang, Wenbin Hu, and Dacheng Tao present graph augmented reasoning in two thousand twenty-five. Small language models lack deep domain knowledge and hallucinate under strain. Extra training is not an option in tight compute settings. Their answer: weave knowledge graph lookup into each reasoning step.
The frame, KG-RAR, builds process oriented knowledge graphs first. A staged lookup pulls facts as each reasoning step needs them. A post retrieval module cleans facts and scores each step. Frozen small models gain math skill with no new training. Math500 scores rose by a fifth for Llama-3B.
Catalog record
| CREATOR | Wenjie Wu, Yongcheng Jing, Yingjie Wang, Wenbin Hu, Dacheng Tao |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-01642