Han Li

WeLM Team, Weixin AI

Han Li's Google Scholar profile photo

About

I am a researcher on the WeLM team at Weixin AI, where I work on post-training for multimodal foundation models. My current research interests include multimodal mid-training and reinforcement learning for general-purpose multimodal models and agents.

Previously, I worked on unified models for multimodal generation and understanding during a research internship with Meituan’s LongCat foundation model team. I received both my bachelor’s degree and Ph.D. from Shanghai Jiao Tong University (SJTU). My Ph.D. advisor was Prof. Hongkai Xiong.

Google Scholar · GitHub ·

Research Interests

  • Unified multimodal models for generation and understanding
  • Multimodal mid-training and foundation model post-training
  • Reinforcement learning for general-purpose multimodal models and agents

Selected Publications

Full list on Google Scholar
  1. 2026 WeAgent-MMGenEdit agentic reinforcement learning framework

    WeAgent-MMGenEdit: A Full-Stack Recipe for Multimodal Agentic Image Generation and Editing

    WeLM Team

    Technical report, 2026 · paper · project

  2. 2026 WeAgent-MMSearch harness and asynchronous reinforcement learning framework

    WeAgent-MMSearch: Native Text-Vision Interaction for Multimodal Search Agents

    WeLM Team

    Technical report, 2026 · paper · project

  3. 2026 OneCAT model and image understanding and generation examples

    OneCAT: Decoder-Only Auto-Regressive Model for Unified Understanding and Generation

    Han Li, Xinyu Peng, Yaoming Wang, Zelin Peng, Xin Chen, Rongxiang Weng, Jingang Wang, Xunliang Cai, Wenrui Dai, Hongkai Xiong

    CVPR 2026 · paper

  4. 2026 Overview of the H2SVC video cache framework

    H2SVC: Head-aware Heterogeneous Streaming Video Cache for Online Video Understanding

    Han Li, Xinyi Zhang, Wenrui Dai, Yaoming Wang, Xinyu Peng, Fan He, Hang Xu, Ziyang Zheng, Chenglin Li, Junni Zou, Hongkai Xiong

    ECCV 2026 · paper · publisher

  5. 2025 Overview of the disentangled image compression model

    On Disentangled Training for Nonlinear Transform in Learned Image Compression

    Han Li, Shaohui Li, Wenrui Dai, Maida Cao, Nuowen Kan, Chenglin Li, Junni Zou, Hongkai Xiong

    ICLR 2025 (Spotlight) · paper

  6. 2024 Architecture of the frequency-aware image compression transformer

    Frequency-Aware Transformer for Learned Image Compression

    Han Li, Shaohui Li, Wenrui Dai, Chenglin Li, Junni Zou, Hongkai Xiong

    ICLR 2024 · paper

Work Experience

2026–present
WeLM Team, Weixin AI
Multimodal foundation model post-training
2024–2026
LongCat Team
Research intern; unified multimodal generation and understanding

Education

2021–2026
Shanghai Jiao Tong University
Ph.D.
2017–2021
Shanghai Jiao Tong University
Bachelor’s degree