About me 关于我
I am a Ph.D. student in Artificial Intelligence at Zhejiang University, where I am a member of REAL Lab, advised by Dr. Yongliang Shen. I expect to graduate around 2027. Before that, I received my B.S. from Beijing Normal University.
My research centers on large language model (LLM) reasoning and AI agents, with a focus on how to make reasoning models think more effectively and efficiently. My work spans long-horizon and infinite-horizon reasoning, reinforcement learning for reasoning, reward modeling and verification, and agents that act reliably in real environments. To date I have published 22 papers at top venues including NeurIPS, ICML, ICLR, ACL, EMNLP, CVPR, ECCV and AAAI, 9 of them as first or co-first author.
Alongside my Ph.D., I have worked as an intern on foundation-model teams in industry, including Meituan (LongCat), Ant Group (Ling / Ring), and currently the Tencent Hunyuan VLM Post-Training Team, on large-scale pre-training, post-training, and reinforcement learning for reasoning models.
You can find my work on Google Scholar, Semantic Scholar, and OpenReview. Feel free to reach out at yanyuchen@zju.edu.cn if you would like to discuss research or potential collaborations.
Research Interests
- LLM Reasoning: long-horizon and infinite-horizon reasoning, efficient and adaptive reasoning, chain-of-thought tuning
- Reinforcement Learning for LLMs: reward modeling, verification, policy optimization
- AI Agents: tool use, GUI and mobile agents, agents grounded in environments
News
- 2026.08 GSM8K-V (co-first author) and IterSynth accepted to EMNLP 2026 Main; Code-A1 (co-first author) and LAPO to Findings; EasySteer to Demo.
- 2026.07 ViewSpatial-Bench and Learning from Reliable Negatives accepted to ECCV 2026.
- 2026.06 InftyThink+ and Milestone-Guided Policy Learning accepted to ICML 2026.
- 2026.04 Joined the Tencent Hunyuan VLM Post-Training Team as an intern.
- 2026.01 InftyThink, MathFimer, VerifyBench, and SpatialLadder accepted to ICLR 2026.
- 2025.12 Test-Time RL for GUI Grounding accepted to AAAI 2026.
- 2025.09 Mind the Gap and Self-Braking Tuning accepted to NeurIPS 2025 (both co-first author).
我是浙江大学人工智能专业的博士研究生，REAL Lab 成员，导师为沈永亮老师，预计 2027 年前后毕业。本科毕业于北京师范大学。
我的研究聚焦于大模型推理与 AI 智能体，关注如何让推理模型思考得更有效、更高效，具体包括长程与无限视野推理、面向推理的强化学习、奖励建模与验证，以及能在真实环境中可靠行动的智能体。目前已在 NeurIPS、ICML、ICLR、ACL、EMNLP、CVPR、ECCV、AAAI 等顶级会议发表论文 22 篇，其中一作/共同一作 9 篇。
读博期间，我先后在美团（LongCat）、蚂蚁集团（Ling / Ring）以及目前所在的腾讯混元基座大模型团队实习，从事推理模型的大规模预训练、后训练与强化学习工作。
我的论文可以在 Google Scholar、Semantic Scholar 和 OpenReview 上找到。欢迎通过 yanyuchen@zju.edu.cn 联系我，交流研究或探讨合作。
研究方向
- 大模型推理：长程与无限视野推理、高效与自适应推理、思维链微调
- 大模型强化学习：奖励建模、验证、策略优化
- AI 智能体：工具调用、GUI 与移动端智能体、环境交互智能体
动态
- 2026.08 GSM8K-V（共同一作）与 IterSynth 被 EMNLP 2026 主会录用；Code-A1（共同一作）与 LAPO 被 Findings 录用；EasySteer 被 Demo 录用。
- 2026.07 ViewSpatial-Bench 与 Learning from Reliable Negatives 被 ECCV 2026 录用。
- 2026.06 InftyThink+ 与 Milestone-Guided Policy Learning 被 ICML 2026 录用。
- 2026.04 加入腾讯混元 VLM 后训练团队实习。
- 2026.01 InftyThink、MathFimer、VerifyBench 与 SpatialLadder 被 ICLR 2026 录用。
- 2025.12 Test-Time RL for GUI Grounding 被 AAAI 2026 录用。
- 2025.09 Mind the Gap 与 Self-Braking Tuning 被 NeurIPS 2025 录用（均为共同一作）。