Hi, my name is Weize Chen (陈纬泽(现)/陈暐泽(旧)), and I am currently a 4th-year PhD student at Tsinghua University, advised by Prof. Zhiyuan Liu. I am also a proud member of the THUNLP group. I earned my bachelor’s degree from the Department of Computer Science and Technology at Tsinghua University.

My research focuses on natural language processing (NLP) and machine learning (ML), with a particular emphasis on improving the performance and efficiency of agent systems and large language model (LLM) systems.

Specifically, my research interests lie in:

• (Multi-)Agent Systems: Designing and implementing agent systems for LLMs and MLLMs to enhance performance. My work also explores improving agent communication and cooperation for more effective collaboration.
• Reinforcement Learning: Investigating the mechanism of LLM RL, and pushing the limit of LLM RL.

Before ChatGPT, my research interests included neural ODE/SDE, hyperbolic neural networks, and knowledge graphs. These experiences continue to inform my perspective on advancing AI.

News

• [2025-09] ✨ The DIET has been accepted at NeurIPS 2025! We integrate an on-the-fly difficulty estimation to RL to mitigate the overthinking problem of the reasoning LLM.

• [2025-09] 🥳 We have just released a blog (From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones) showing that RL do teach the LLM new generalizable compositional skill, and these skills are transferrable across domains. Check it out!

• [2025-02] 🎉 IoA has been accepted as a spotlight at ICLR 2025!

• [2024-10] 🚀 Optima (Optimizing the effectiveness and efficiency of multi-agent systems) is now released! Our paper demonstrates that iterative training for multi-agent systems significantly enhances performance and efficiency, achieving up to 90% reduction in inference tokens while delivering superior performance. Moreover, with fewer inference tokens, we establish a better inference scaling law!

• [2024-07] 🚀 IoA (Internet of Agents) has been released! We introduce the concept of connecting heterogeneous and distributed agents via the Internet and present a prototype implementation functioning as an instant-messaging app for agents. Our experiments on complex agent tasks, embodied agents, and retrieval-augmented generation (RAG) demonstrate the effectiveness of IoA.

• [2024-01] 🎉 AgentVerse has been accepted to ICLR 2024! See you in Vienna! Also take a try at our GitHub repo, which now has more than 4k stars!

Highlighted Publications:

Please refer to publication for the full list.

The Overthinker’s DIET: Cutting Token Calories with DIfficulty-AwarE Training
Weize Chen*, Jiarui Yuan*, Tailin Jin, Ning Ding, Huimin Chen, Zhiyuan Liu, Maosong Sun
NeurIPS 2025 [paper] [code]

From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones
Lifan Yuan*, Weize Chen*, Yuchen Zhang, Ganqu Cui, Hanbin Wang, Ziming You, Ning Ding, Zhiyuan Liu, Maosong Sun, Hao Peng
Blog Post [blog] [tweet]

Optima: Optimizing Effectiveness and Efficiency for LLM-Based Multi-Agent System
Weize Chen*, Jiarui Yuan*, Chen Qian, Cheng Yang, Zhiyuan Liu, Maosong Sun
ALC 2025 Findings [project page] [pdf] [code]

Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence
Weize Chen*, Ziming You*, Ran Li*, Yitong Guan*, Chen Qian, Chenyang Zhao, Cheng Yang, Ruobing Xie, Zhiyuan Liu, Maosong Sun
ICLR 2025 Spotlight [pdf] [code]

AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
Weize Chen*, Yusheng Su*, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chi-Min Chan, Heyang Yu, Yaxi Lu, Yi-Hsin Hung, Chen Qian, Yujia Qin, Xin Cong, Ruobing Xie, Zhiyuan Liu, Maosong Sun, Jie Zhou
ICLR 2024 [pdf] [code]

ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate
Chi-Min Chan, Weize Chen, Yusheng Su, Jianxuan Yu, Wei Xue, Shanghang Zhang, Jie Fu, Zhiyuan Liu
ICLR 2024 [pdf] [code]

Fully Hyperbolic Neural Networks
Weize Chen*, Xu Han*, Yankai Lin, Hexu Zhao, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou
ACL 2022 [pdf] [code]

Quantifying Similarity between Relations with Fact Distribution
Weize Chen, Hao Zhu, Xu Han, Zhiyuan Liu, Maosong Sun
ACL 2019 Oral [pdf] [code]