About
Learning to reason,
efficiently.
I study how large language models can learn stronger reasoning abilities from less data and computation.
Biography
I am a Ph.D. student in Computer Science at Clemson University, where I work in the CLAIM Lab under the supervision of Prof. Kunpeng Liu.
Previously, I received my master's and bachelor's degrees from Tsinghua University. During my master's study, I worked in the Multimedia Signal and Intelligent Information Processing Lab with Prof. Ji Wu. My recent work focuses on making post-training for large reasoning models more adaptive, stable, predictable, and data-efficient.
- Large Language Models
- Reinforcement Learning
- Self-Distillation
- Data-Efficient Learning
- Reasoning Models
News
- DualOPSD, our work on adaptive privileged teachers for on-policy self-distillation, was released on arXiv.
- Joined Clemson University as a Ph.D. student and received the Robert M. Geist Endowed Fellowship.
- CoScale-RL was released on arXiv.
- Released our study on small-scale fine-tuning in R1-style reinforcement learning, together with code.
- Our work on controllable, speech-driven facial animation appeared at ICMEW 2023.
Education
Ph.D. in Computer Science
Clemson University · 2026–present · Advisor: Kunpeng Liu
Master of Electronic and Information Engineering
Tsinghua University · 2023–2026 · Advisor: Ji Wu
Bachelor of Electronic Information Science and Technology
Tsinghua University · 2019–2023