Agentic Reasoning: Reasoning LLMs with Tools for the Deep Research
Junde Wu, Jiayuan Zhu, Yuyuan Liu · 2025
Read Agentic Reasoning: Reasoning LLMs with Tools for the Deep Research by Junde Wu, Jiayuan Zhu, Yuyuan Liu (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-2502-04644.txt.
Description
Junde Wu and co-authors build a framework that lets a language model call on outside tools while it reasons. Their 2025 paper targets hard questions that need deep research. The framework adds three helpers. One searches the web, one runs code, and one keeps a structured map of the reasoning so far. The map stores facts and links between them, which keeps long chains of thought coherent.
The search helper gets special care and beats prior search designs in their tests. Plugged into a strong open model, the full setup reaches the top score among public models. Its results rival a leading closed research product. Tests that remove one helper at a time confirm each part earns its place. The work shows tool use and memory maps can lift machine reasoning a long way.
Catalog record
| CREATOR | Junde Wu, Jiayuan Zhu, Yuyuan Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-04644