Senior Researcher
Microsoft Research AI for Science

Google Scholar | Github

guoqingliu (at) microsoft.com
21 Station Road, Cambridge, CB1 2FB, United Kingdom

• 2025-02: Our paper HybriDNA: A Hybrid Transformer-Mamba2 Long-Range DNA Language Model is available at arXiv!
• 2025-02: Our paper NatureLM: Deciphering the Language of Nature for Scientific Discovery is available at arXiv!
• 2024-12: Our paper Chimera: Accurate retrosynthesis prediction by ensembling models with diverse inductive biases is available at arXiv!
• 2024-07: Our paper Reinforcement Learning from Bagged Reward is accepted to TMLR!
• 2024-05: Our paper Token-level Direct Preference Optimization is accepted to ICML 2024!
• 2024-01: Our paper Connecting Large Language Models with Evolutionary Algorithms Yields Powerful Prompt Optimizers is accepted to ICLR 2024!
• 2023-10: Our paper De novo Drug Design using Reinforcement Learning with Multiple GPT Agents is accepted to NeurIPS 2023!
• 2023-04: Our paper Retrosynthetic Planning with Dual Value Networks is accepted to ICML 2023!

Guoqing Liu is a Senior Researcher at Microsoft Research AI for Science. His research interests lie in reinforcement learning (RL), intersection of foundation models and decision making/RL, and their applications in scientific discovery. He has extensive experience in the research and application of deep RL. Notably, in Project Suphx, he and his team developed Suphx, a world-class Mahjong AI that achieved 10 DAN status on the Tenhou platform in mid-2019. In Project Mariana, he led the development of Inspector, the first general pixel-based automated game testing agent, which was deployed by Xbox Studios Quality. Recently, his work has centered on exploring the applications of large language/foundation models and RL agents in drug discovery (e.g., PDVN, RetroChimera, NaturLM). Before joining Microsoft Research, he earned his Ph.D. degree from University of Science and Technology of China (USTC) in 2021 through a joint Ph.D. program between USTC and Microsoft Research Asia (MSRA), under the supervision of Dr. Tie-Yan Liu and Prof. Nenghai Yu.

("*": equal contribution; "†": correspondence)
1. HybriDNA: A Hybrid Transformer-Mamba2 Long-Range DNA Language Model [Paper]
   Mingqian Ma*, Guoqing Liu*, Chuan Cao*, Pan Deng*, Tri Dao, Albert Gu, Peiran Jin, Zhao Yang, Yingce Xia, Renqian Luo, Pipi Hu, Zun Wang, Yuan-Jyue Chen, Haiguang Liu, Tao Qin. arXiv 2025.
   
2. NatureLM: Deciphering the Language of Nature for Scientific Discovery [Paper]
   NatureLM Team, Microsoft Research AI for Science. arXiv 2025.
   
3. 3DMolFormer: A Dual-channel Framework for Structure-based Drug Discovery [Paper]
   Xiuyuan Hu, Guoqing Liu†, Can Chen, Yang Zhao, Hao Zhang, Xue Liu
   Thirteenth International Conference on Learning Representations (ICLR 2025)
4. Chimera: Accurate retrosynthesis prediction by ensembling models with diverse inductive biases [Paper]
   Krzysztof Maziarz*, Guoqing Liu*, Hubert Misztela, Aleksei Kornev, Piotr Gaiński, Holger Hoefling, Mike Fortunato, Rishi Gupta, Marwin Segler. arXiv 2024.
   
5. Reinforcement Learning from Bagged Reward [Paper]
   Yuting Tang, Xin-Qiang Cai, Yao-Xiang Ding, Qiyu Wu, Guoqing Liu, Masashi Sugiyama
   ICML 2024 Workshop on Aligning Reinforcement Learning Experimentalists and Theorists (ICML-W 2024)
6. Token-level Direct Preference Optimization [Paper]
   Yongcheng Zeng, Guoqing Liu, Weiyu Ma, Ning Yang, Haifeng Zhang, Jun Wang
   Forty-first International Conference on Machine Learning (ICML 2024)
7. Connecting Large Language Models with Evolutionary Algorithms Yields Powerful Prompt Optimizers [Paper]
   Qingyan Guo, Rui Wang, Junliang Guo, Bei Li, Kaitao Song, Xu Tan, Guoqing Liu, Jiang Bian, Yujiu Yang
   Twelfth International Conference on Learning Representations (ICLR 2024)
8. Re-evaluating Retrosynthesis Algorithms with Syntheseus [Paper]
   Krzysztof Maziarz, Austin Tripp, Guoqing Liu, Megan Stanley, Shufang Xie, Piotr Gaiński, Philipp Seidl, Marwin Segler
   NeurIPS 2023 Workshop on AI for Science (NeurIPS-W 2023)
9. De novo Drug Design using Reinforcement Learning with Multiple GPT Agents [Paper]
   Xiuyuan Hu, Guoqing Liu†, Yang Zhao, Hao Zhang
   Thirty-seventh Conference on Neural Information Processing Systems (NeurIPS 2023)
10. QuinNet: Efficiently Incorporating Quintuple Interactions into Geometric Deep Learning Force Fields [Paper]
    Zun Wang, Guoqing Liu, Yichi Zhou, Tong Wang, Bin Shao
    Thirty-seventh Conference on Neural Information Processing Systems (NeurIPS 2023)
11. Retrosynthetic Planning with Dual Value Networks [Paper]
    Guoqing Liu*, Di Xue*, Shufang Xie, Yingce Xia, Austin Tripp, Krzysztof Maziarz, Marwin Segler, Tao Qin, Zongzhang Zhang, Tie-Yan Liu
    Fortieth International Conference on Machine Learning (ICML 2023)
12. Re-evaluating Chemical Synthesis Planning Algorithms [Paper]
    Austin Tripp, Krzysztof Maziarz, Sarah Lewis, Guoqing Liu, Marwin Segler
    NeurIPS 2022 Workshop on AI for Science (NeurIPS-W 2022)
13. You May Not Need Ratio Clipping in PPO [Paper]
    Mingfei Sun, Vitaly Kurin, Guoqing Liu, Sam Devlin, Tao Qin, Katja Hofmann, Shimon Whiteson. arXiv 2022.
    
14. Inspector: Pixel-based Automated Game Testing via Exploration, Detection, and Investigation [Paper]
    Guoqing Liu, Mengzhang Cai, Li Zhao, Tao Qin, Adrian Brown, Jimmy Bischoff and Tie-Yan Liu
    IEEE Conference on Games 2022 (COG 2022, Oral)
15. Independence-aware Advantage Estimation [Paper]
    Pushi Zhang, Li Zhao, Guoqing Liu, Jiang Bian, Minlie Huang, Tao Qin, Tie-Yan Liu
    30th International Joint Conference on Artificial Intelligence (IJCAI 2021)
16. Return-based Contrastive Representation Learning for Reinforcement Learning [Paper]
    Guoqing Liu*, Chuheng Zhang*, Li Zhao, Tao Qin, Jinhua Zhu, Jian Li, Nenghai Yu, Tie-Yan Liu
    Ninth International Conference on Learning Representations (ICLR 2021)
17. Demonstration Actor Critic [Paper]
    Guoqing Liu, Li Zhao, Pushi Zhang, Jiang Bian, Tao Qin, Nenghai Yu, Tie- Yan Liu
    Neurocomputing, Volume 434, 28 April 2021, Pages 194-202 (Neurocomputing 2021)
18. Suphx: Mastering Mahjong with Deep Reinforcement Learning [Paper][News]
    Junjie Li, Sotetsu Koyamada, Qiwei Ye, Guoqing Liu, Chao Wang, Ruihan Yang, Li Zhao, Tao Qin, Tie-Yan Liu, Hsiao-Wuen Hon. arXiv 2020.
    
19. Trust Region Evolution Strategies [Paper]
    Guoqing Liu, Li Zhao, Feidiao Yang, Jiang Bian, Tao Qin, Nenghai Yu, Tie-Yan Liu
    Thirty-Third AAAI Conference on Artificial Intelligence (AAAI 2019)

• Sep.2016 - Jun.2021, School of Cyberspace Security, University of Science and Technology of China, Doctor of Philosophy (Ph.D.)
• Sep.2012 - Jun.2016, School of Electronic Engineering, University of Science and Technology of China, Bachelor of Engineering (B.E.)

• PC member: ICLR, ICML, NeurIPS, AAAI, AAMAS
• Journal reviewer: TPAMI, Neurocomputing

• Aug.2016, Microsoft Research Asia, Stars of Tomorrow Internship Award
• Sep.2013, Sep.2014, Sep.2015, University of Science and Technology of China, Excellent Student Scholarship of USTC
• Dec.2013, Chinese Mathematical Society (CMS), First Price of National College Students Mathematical Competition
• Oct.2011, Chinese Mathematical Society (CMS), First Price of National High School Mathematics League