Chem-R: Learning to Reason as a Chemist
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911225604997120 |
|---|---|
| author | Wang, Weida Chen, Benteng Zhang, Di Liu, Wanhao Pu, Shuchen Gao, Ben Zeng, Jin Wei, Xiaoyong Yu, Tianshu Sun, Shuzhou Fu, Tianfan Ouyang, Wanli Bai, Lei Li, Jiatong Wang, Zifu Li, Yuqiang Zhang, Shufei |
| author_facet | Wang, Weida Chen, Benteng Zhang, Di Liu, Wanhao Pu, Shuchen Gao, Ben Zeng, Jin Wei, Xiaoyong Yu, Tianshu Sun, Shuzhou Fu, Tianfan Ouyang, Wanli Bai, Lei Li, Jiatong Wang, Zifu Li, Yuqiang Zhang, Shufei |
| contents | Although large language models (LLMs) have significant potential to advance chemical discovery, current LLMs lack core chemical knowledge, produce unreliable reasoning trajectories, and exhibit suboptimal performance across diverse chemical tasks. To address these challenges, we propose Chem-R, a generalizable Chemical Reasoning model designed to emulate the deliberative processes of chemists. Chem-R is trained through a three-phase framework that progressively builds advanced reasoning capabilities, including: 1) Chemical Foundation Training, which establishes core chemical knowledge. 2) Chemical Reasoning Protocol Distillation, incorporating structured, expert-like reasoning traces to guide systematic and reliable problem solving. 3) Multi-task Group Relative Policy Optimization that optimizes the model for balanced performance across diverse molecular- and reaction-level tasks. This structured pipeline enables Chem-R to achieve state-of-the-art performance on comprehensive benchmarks, surpassing leading large language models, including Gemini-2.5-Pro and DeepSeek-R1, by up to 32% on molecular tasks and 48% on reaction tasks. Meanwhile, Chem-R also consistently outperforms the existing chemical foundation models across both molecular and reaction level tasks. These results highlight Chem-R's robust generalization, interpretability, and potential as a foundation for next-generation AI-driven chemical discovery. The code and model are available at https://github.com/davidweidawang/Chem-R. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_16880 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Chem-R: Learning to Reason as a Chemist Wang, Weida Chen, Benteng Zhang, Di Liu, Wanhao Pu, Shuchen Gao, Ben Zeng, Jin Wei, Xiaoyong Yu, Tianshu Sun, Shuzhou Fu, Tianfan Ouyang, Wanli Bai, Lei Li, Jiatong Wang, Zifu Li, Yuqiang Zhang, Shufei Computational Engineering, Finance, and Science Although large language models (LLMs) have significant potential to advance chemical discovery, current LLMs lack core chemical knowledge, produce unreliable reasoning trajectories, and exhibit suboptimal performance across diverse chemical tasks. To address these challenges, we propose Chem-R, a generalizable Chemical Reasoning model designed to emulate the deliberative processes of chemists. Chem-R is trained through a three-phase framework that progressively builds advanced reasoning capabilities, including: 1) Chemical Foundation Training, which establishes core chemical knowledge. 2) Chemical Reasoning Protocol Distillation, incorporating structured, expert-like reasoning traces to guide systematic and reliable problem solving. 3) Multi-task Group Relative Policy Optimization that optimizes the model for balanced performance across diverse molecular- and reaction-level tasks. This structured pipeline enables Chem-R to achieve state-of-the-art performance on comprehensive benchmarks, surpassing leading large language models, including Gemini-2.5-Pro and DeepSeek-R1, by up to 32% on molecular tasks and 48% on reaction tasks. Meanwhile, Chem-R also consistently outperforms the existing chemical foundation models across both molecular and reaction level tasks. These results highlight Chem-R's robust generalization, interpretability, and potential as a foundation for next-generation AI-driven chemical discovery. The code and model are available at https://github.com/davidweidawang/Chem-R. |
| title | Chem-R: Learning to Reason as a Chemist |
| topic | Computational Engineering, Finance, and Science |
| url | https://arxiv.org/abs/2510.16880 |