Ph.D. Student, HKUST CSE Research Intern, Tencent Hy
I am a Ph.D. student in the Department of Computer Science and Engineering at the Hong Kong University of Science and Technology (HKUST), advised by Prof. Kai Chen in the iSING Lab. Previously, I received my M.Phil. degree from HKUST CSE, also advised by Prof. Kai Chen, and my B.S. degree in Applied Mathematics from Northwestern Polytechnical University.
I am currently a research intern on the RL Team of the Foundation Model Department at Tencent Hy (formerly Tencent Hunyuan), as part of the Tencent Qingyun Program. My research focuses on long-horizon agent reinforcement learning, as well as algorithms and infrastructure for agent RL training.
Tencent Hy · arXiv:2608.29296, 2026
A larger batch shortens wall-clock time-to-target only when its throughput gain exceeds its samples-to-target penalty. In GRPO, pairing larger batches with learning-rate retuning reduces time-to-target by up to 29%.
Conference on Neural Information Processing Systems (NeurIPS 2026), Main Conference
PaperSoon CodeSoon
ACM SIGOPS Annual Technical Conference (ATC 2026)
PaperSoon CodeSoon
International Conference on Learning Representations (ICLR 2026)
Conference on Language Modeling (COLM 2026)
arXiv:2609.10226, 2026 · * Equal contribution
Open-source Tencent Hy models I have contributed to.
295B MoE · 21B active
295B MoE · 21B active
770B MoE · 49B active
RL Team, Foundation Model Department · Tencent Qingyun Program
Working on long-horizon agent RL and on algorithms and infrastructure for agent RL training.
The Hong Kong University of Science and Technology · Advisor: Prof. Kai Chen
The Hong Kong University of Science and Technology · Advisor: Prof. Kai Chen
Northwestern Polytechnical University