Huayu Chen (陈华玉)  PhD student Room 1-509, FIT Building Dept. of Computer Science and Technology Tsinghua University Beijing, China, 100084. Email: chenhuay17[AT]gmail[DOT]com [Google Scholar]

Biography

I am a fourth-year PhD Candidate of TSAIL Group in the Department of Computer Science and Technology, Tsinghua University, advised by Prof. Jun Zhu and Prof. Hang Su. Currently, I am also a research intern at Nvidia Deep Imagination Research group in the San Francisco Bay Area, working with Dr. Ming-yu Liu. Previously, I received my B.S. degree from the Department of Automation of Tsinghua University in July 2021. I spent a wonderful time at the Digital Media Lab at Tsinghua University, advised by Prof. Yebin Liu in the field of AIGC from Oct 2018 to May 2019. I have also been a research intern at Netease's Fuxi AI Lab and ByteDance's AI Lab respectively in 2021 and 2020.

Currently, my research interests lie primarily in the area of deep reinforcement learning and deep generative models. My lifelong goal is to build a scalable, impenetrable, and adaptable decision-making engine that could relieve human from tedious tasks. My current progress includes authoring Tianshou: A highly modularized deep reinforcement learning library , designing large-scale Online RL system for mastering MOBA games (see Competitions), and bridging the gap between RL theories and generative models such as diffusion/LLM. Currently, I aim at unifying the SL/RL paradigm for different domains such as vision and language.

Looking for academic and industrial research positions. Expect to graduate in June 2026. My CV is [here]. Drop me an email if you feel interested.

Selected Publications

RL Infra:

• Tianshou: A Highly Modularized Deep Reinforcement Learning Library
  Jiayi Weng*, Huayu Chen*, Dong Yan, Kaichao You, Alexis Duburcq, Minghao Zhang, Yi Su, Hang Su, Jun Zhu
  Journal of Machine Learning Research (JMLR)
  [Code: 8.5k Stars]

Vision RL (Diffusion & AR):

• Visual Generation Without Guidance
  Huayu Chen*, Kai Jiang*, Kaiwen Zheng, Jianfei Chen, Hang Su, Jun Zhu
  International Conference on Machine Learning (ICML 2025)
  [code]
• Toward Guidance-Free AR Visual Generation via Condition Contrastive Alignment
  Huayu Chen, Hang Su, Peize Sun, Jun Zhu
  International Conference on Learning Representations (ICLR 2025)
  Oral (Accept rate~1.8%)
  [code]
• Contrastive Energy Prediction for Exact Energy-Guided Diffusion Sampling in Offline Reinforcement Learning
  Cheng Lu*, Huayu Chen*, Jianfei Chen, Hang Su, Chongxuan Li, Jun Zhu
  International Conference on Machine Learning (ICML 2023)
  [code] [poster]
• Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator
  Kaiwen Zheng, Yongxin Chen, Huayu Chen, Guande He, Ming-Yu Liu, Jun Zhu, Qinsheng Zhang
  International Conference on Machine Learning (ICML 2025)
  Spotlight (Accept rate~2.6%)
  

Language/Multimodal RL (LLM & VLM):

• Bridging Supervised Learning and Reinforcement Learning in Math Reasoning
  Huayu Chen, Kaiwen Zheng, Qinsheng Zhang, Ganqu Cui, Yin Cui, Haotian Ye, Tsung-Yi Lin, Ming-Yu Liu, Jun Zhu, Haoxiang Wang
  [Project Page] [Preprint]
• Cosmos-Reason1: From Physical Common Sense To Embodied Reasoning
  Nvidia Group (Core contributor, for VLM RL training)
  [project page] [code]
• Noise Contrastive Alignment of Language Models with Explicit Rewards
  Huayu Chen, Guande He, Lifan Yuan, Ganqu Cui, Hang Su, Jun Zhu
  Annual Conference on Neural Information Processing Systems (NeurIPS 2024)
  [code]
• Free Process Rewards without Process Labels
  Lifan Yuan, Wendi Li, Huayu Chen, Ganqu Cui, Ning Ding, Kaiyan Zhang, Bowen Zhou, Zhiyuan Liu, Hao Peng
  (Propose the ImplicitPRM algorithm employed by PRIME [1.6k Stars])
  International Conference on Machine Learning (ICML 2025)
  [code]

Embodied RL (Diffusion Policy):

• RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation
  Songming Liu, Lingxuan Wu, Bangguo Li, Hengkai Tan, Huayu Chen, Zhengyi Wang, Ke Xu, Hang Su, Jun Zhu
  International Conference on Learning Representations (ICLR 2025)
  [project page] [code]
• Aligning Diffusion Behaviors with Q-functions for Efficient Continuous Control
  Huayu Chen, Kaiwen Zheng, Hang Su, Jun Zhu
  Annual Conference on Neural Information Processing Systems (NeurIPS 2024)
  
• Score Regularized Policy Optimization through Diffusion Behavior
  Huayu Chen, Cheng Lu, Zhengyi Wang, Hang Su, Jun Zhu
  International Conference on Learning Representations (ICLR 2024)
  [code] [poster]
• Offline Reinforcement Learning via High-Fidelity Generative Behavior Modeling
  Huayu Chen, Cheng Lu, Chengyang Ying, Hang Su, Jun Zhu
  International Conference on Learning Representations (ICLR 2023)
  [code] [poster]

* indicates co-first authors.

Competitions

• First place (two years in a row) in Tencent's multi-agent RL competition of Honor of Kings (王者荣耀), final win rate: 99.2% , 2021-2023
  [news] [webpage]
• Second place in DJI's Robomaster Sim2Real Challenge, ICRA 2022
• First place in the 30th International Design Contest (IDC Robocon 2019, MIT), 2019
• First place in the 20th Electronic Design Competition at Tsinghua University, 2018
• First place in the 1st Artificial Intelligence Challenge at Tsinghua University, 2017

Honors & Awards

• HUAWEI-Tsinghua Scholarship, 2023
• '84' Future Innovation Scholarship, 2023
• Outstanding Undergraduate in Beijing, 2021
• BaoGang Scholarship (Awarded to ~500 students in China every year), 2021
• Student Of The Year, in Dept. of Automation, Tsinghua University, 2020
• China National Scholarship, 2019
• Excellence Award for Technological Innovation, Tsinghua University, 2019
• '129' Scholarship (Highest honor for 2nd year students at Tsinghua), 2018
• 1st Prize in the 35th China Regional College Students Physics Competition, 2018
• 1st Prize in the 30th National Physics Olympiad, 2016

Services

Reviewer for ICLR, NeurIPS, ICML, AISTATS, AAAI, etc.

President of Student Association of Science and Technology, Dept. of Automation, Tsinghua University, 2020-2021

Teaching

2023 Spring, TA in Statistical Learning Theory and Applications, instructed by Prof. Jun Zhu