👀 About Me
Hello! I am a master’s student majoring in software engineering in the Chinese Information Processing Laboratory at the Institute of Software Chinese Academy of Sciences, co-advised by Prof. Xianpei Han and Assoc. Prof. Yaojie Lu.
My research focuses on realistic user behavior simulation and reliable multi-turn interactions in LLM-based agents. I am particularly interested in understanding how users make decisions across real-world scenarios and how these insights can help evaluate and improve interactive agents, especially in commercial applications.
At Kuaishou, I work on user simulation and post-training for LLM-based customer service assistants, connecting behavior modeling with practical challenges in business applications. As first author, I have led three closely connected research projects: OmniBehavior, a benchmark grounded in real user interaction traces for evaluating long-horizon behavior simulation; OmniSimulator, which trains language models to simulate diverse user behaviors across business scenarios; and OASIS, a framework for improving customer service assistants through learning from users’ latent thoughts, with the goal of increasing conversion and user satisfaction. Together, these projects address how to evaluate user simulators, train them, and use insights into user behavior to improve multi-turn interactions.
I also led the research on ConsistentChat to improve dialogue consistency over long multi-turn conversations, and contributed to ZServe, an adaptive framework for efficient LLM serving in production environments.
🔥 News
- 2026.09: 🎉 OmniBehavior and OmniSimulator were accepted by NeurIPS 2026! Big thanks to all my co-authors for their contributions!
- 2026.08: 🎉 One paper was accepted by EMNLP 2026!
- 2026.05: 🔥 We released the complete OmniBehavior dataset and code! Please check them out and feel free to use them in your research.
- 2026.05: 🎉 Our LiveMCPBench was accepted by KDD 2026!
- 2026.04: 🎉 Our CoCoNUTS was accepted by ACL 2026! Our MemSearcher was accepted by Findings of ACL 2026!
- 2026.04: 🔥 Our OmniBehavior paper was released! The first open-source benchmark based on real users, featuring long-horizon, cross-commercial-scenario, heterogeneous behavior traces of Kuaishou platform.
- 2025.10: 🎉 Our ConsistentChat was selected as Oral presentation at EMNLP 2025! See you in Suzhou!
- 2025.08: 🎉 Our ConsistentChat was accepted by EMNLP 2025!
- 2025.06: 🎉 Our ShortV was accepted by ICCV 2025!
📝 Publications
Towards Real-world Human Behavior Simulation: Benchmarking Large Language Models on Long-horizon, Cross-scenario, Heterogeneous Behavior Traces
Jiawei Chen, Ruoxi Xu, Boxi Cao, Ruotong Pan, Yunfei Zhang, Yifei Hu, Yong Du, Tingting Gao, Yaojie Lu, Yingfei Sun, Xianpei Han, Le Sun, Xiangyu Wu, Hongyu Lin
首个基于真实用户的长序列、跨场景、异构的大模型行为模拟基准。
NeurIPS 2026
[Paper] [Code] [Project Page] [Dataset] [小红书]
OmniSimulator: Aligning Small Language Models for Authentic Heterogeneous Behavior Modeling
Jiawei Chen, Ruoxi Xu, Boxi Cao, Zhuoqun Li, Ruotong Pan, Yunfei Zhang, Zhirui Yang, Jingyan Chen, Tingting Gao, Han Li, Yaojie Lu, Xianpei Han, Le Sun, Xiangyu Wu, Hongyu Lin
基于真实用户行为训练的通用用户模拟器，在7B的小模型上实现了超越 Claude-Sonnet-4.5 的行为模拟效果。
NeurIPS 2026
Learning from What Users Do Not Say: Self-Evolving Assistants by Internalizing Latent Thoughts
Jiawei Chen, Ruotong Pan, Boxi Cao, Yunfei Zhang, Lingdi Kong, Wenwu Ou, Xianpei Han, Le Sun, Xiangyu Wu, Hongyu Lin
从用户未显式表达的潜在意图与思考中学习实现模型自进化，帮助下游智能客服持续优化多轮交互策略与响应决策。
View Full Publication List Hide Full Publication List
LiveMCPBench: Can Agents Navigate an Ocean of MCP Tools?
Guozhao Mo, Wenliang Zhong, Jiawei Chen, Qianhao Yuan, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Le Sun
KDD 2026
[Paper] [Code] [Project Page] [Huggingface]
PraMem: Practice-derived Experiential Memory for Long-horizon Behavior Prediction
Zhuoqun Li, Boxi Cao, Jiawei Chen, Hanshu Zhou, Ruoxi Xu, Guiping Jiang, Ruotong Pan, Tingting Gao, Han Li, Xiangyu Wu, Hongyu Lin, Yaojie Lu, Xianpei Han, Le Sun
[Paper]
ShortV: Efficient Multimodal Large Language Models by Freezing Visual Tokens in Ineffective Layers
Qianhao Yuan, Qingyu Zhang, Yanjiang Liu, Jiawei Chen, Yaojie Lu, Hongyu Lin, Jia Zheng, Xianpei Han, Le Sun
ICCV 2025
[Paper]
MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning
Qianhao Yuan, Jie Lou, Zichao Li, Jiawei Chen, Yaojie Lu, Hongyu Lin, Le Sun, Debing Zhang, Xianpei Han
ACL 2026 Findings
[Paper]
Navigating the Infinite Dynamic Web Space: Effective In-Context Exploration via Cognitive Multi-Agent Collaboration
Guozhao Mo, Yanjiang Liu, Yafei Shi, Jiawei Chen, Yang Li, Yaojie Lu, Hongyu Lin, Ben He, Le Sun, Bo Zheng, Xianpei Han
EACL 2026
[Paper]
ZServe: An Adaptive Framework for Optimizing Large Language Models Serving in Heterogeneous Infrastructures
Taosong Fang, Zhengzhao Ma, Jiawei Chen, Yushen Yang, Kaijie Zhou, Jia Zheng, Xianpei Han, Le Sun, Hongyu Lin, Yaojie Lu
EMNLP 2026 Demo
[Paper]
Beyond Surface Style: Aligning Multi-Turn User Simulators with Behavioral Consistency
Geng Chen, Ruotong Pan, Zhirui Yang, Qiqi He, Jiawei Chen, Zhang Yunfei, Chongyuan Chen, Minxuan Lv, Zheng Yang, Win-Bin Huang, Xiangyu Wu, Wenwu Ou
[Paper]
🎖 Honors and Awards
- 2026.10 National Scholarship (研究生国家奖学金), Ministry of Education of China.
- 2026.05 Outstanding Student Leader (优秀学生干部), University of Chinese Academy of Sciences.
- 2025.11 Highlight poster of 4th National Large Model and Generation Conference, Chinese Information Processing Society.
- 2024.06 Outstanding Graduate named by Zhejiang (省优秀毕业生).
- 2024.06 Presidential Scholarship (校长奖学金).
- 2023.10 National Scholarship (本科生国家奖学金), Ministry of Education of China.
- 2023.06 Jack Ma Youth Leader Award (马云青春领袖奖·十佳大学生).
💻 Internships
- 2025.09 - Present, LLM Research Intern, Foundation Model and Application Team, Kuaishou Technology, China.
- Adviser: Xiangyu Wu
- Research Focus: User agent for business scenarios, post-training and evaluation.
- 2023.10 - 2024.08, Research Intern, Institute of Software, Chinese Academy of Sciences, China.
- Adviser: Yaojie Lu
- Research Focus: research on enhancing LLMs’ ability through RAG, SFT, RLHF ,etc.