About Me

I am an incoming M.S. student at Peking University, advised by Prof. Li Yuan. I have also been fortunate to work closely with Prof. Lei Feng and Prof. Ziwei Liu.

My core research focuses on visual representations and language priors. Vision directly depicts the physical appearance of the "real world"; while language, as the refinement and abstraction of visual information, constructs the knowledge landscape of the "human world." I firmly believe that bridging the gap between the language modality, which carries "human world" prior knowledge, and the visual modality, which directly reflects the "real world," is a crucial step towards achieving general artificial intelligence. This not only involves finding effective mapping paths between modalities but also concerns how to leverage their respective strengths to build a unified framework collaboratively.

Currently, I am dedicated to building native unified multimodal models.

🤝 I am eager to discuss potential collaborations and am actively seeking industry internship opportunities in multimodal models. Please feel free to contact me via email or WeChat: purshow if you are interested.

Education

Experience

Publications

Selected Publications & Manuscripts 6

Full paper list can be found on Google Scholar. * Equal Contribution    † Project Lead

  • WISE teaser
    WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation
    Yuwei Niu*, Munan Ning*, Mengren Zheng, Weiyang Jin, Bin Lin, Peng Jin, Jiaqi Liao, Chaoran Feng, Kunpeng Ning, Bin Zhu, Li Yuan†.
    Paper Code ICML 2026
  • UniSandBox teaser
    Does Understanding Inform Generation in Unified Multimodal Models? From Analysis to Path Forward
    Yuwei Niu*, Weiyang Jin*, Jiaqi Liao, Chaoran Feng, Peng Jin, Bin Lin, Zongjian Li, Bin Zhu, Weihao Yu, Li Yuan†.
    Paper Code Under Review
  • LangBridge teaser
    LangBridge: Interpreting Image as a Combination of Language Embeddings
    Jiaqi Liao*, Yuwei Niu*, Fanqing Meng*, Hao Li, Changyao Tian, Yinuo Du, Yuwen Xiong, Dianqi Li, Xizhou Zhu, Li Yuan, Jifeng Dai†, Yu Cheng†.
    Paper Code ICCV 2025
  • SRUM teaser
    SRUM: Fine-Grained Self-Rewarding for Unified Multimodal Models
    Weiyang Jin*, Yuwei Niu*, Jiaqi Liao, Chengqi Duan, Aoxue Li, Shenghua Gao, Xihui Liu†.
    Paper Code ECCV 2026
  • BDetCLIP teaser
    Test-Time Multimodal Backdoor Detection by Contrastive Prompting
    Yuwei Niu*, Shuo He*, Qi Wei, Zongyu Wu, Feng Liu, Lei Feng†.
    Paper Code ICML 2025
  • Look-Back teaser
    Look-Back: Implicit Visual Re-focusing in MLLM Reasoning
    Shuo Yang*, Yuwei Niu*, Yuyang Liu†, Yang Ye, Bin Lin, Li Yuan†.
    Paper Code AAAI 2026
Projects & Tech Report 6
  • SenseNova-U1 teaser
    SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture
    SenseNova-U1 Team
    Paper Code Stars Tech Report
  • NEO-ov teaser
    From Pixels to Words -- Towards Native One-Vision Models at Scale
    NEO-ov Team
    Paper Code Stars Tech Report
  • UniWorld teaser
    UniWorld: High-Resolution Semantic Encoders for Unified Visual Understanding and Generation
    UniWorld Team
    Paper Code Stars Tech Report
  • UniWorld-V2 teaser
    Uniworld-V2: Reinforce Image Editing with Diffusion Negative-aware Finetuning and MLLM Implicit Feedback
    UniWorld Team
    Paper Code Stars Tech Report
  • lmms-eval teaser
    LMMs-Eval: Probing Intelligence in the Real World
    LMMs-Lab
    Code Stars Codebase
  • lmms-engine teaser
    LMMS-Engine: A simple, unified multimodal models training engine. Lean, flexible, and built for hacking at scale.
    LMMs-Lab
    Code Stars Codebase

Blog

I occasionally write blogs to share my thoughts. Chinese readers can find the full archive on my RedNote.

Personality

Beyond science, I am deeply passionate about literature, poetry, anime, films, music, and all forms of art that embody human creativity. I firmly believe these are the reasons for our existence. Currently, I am deeply captivated by the literature of Dostoevsky and Tolstoy, and R&B music (such as Prince, Stevie Wonder, D'Angelo, David Tao, and Khalil Fong).

Contact Me

I believe good ideas and meaningful progress often come from open discussion and thoughtful disagreement. If you have comments on my work, different perspectives, or new ideas you would like to share, I would be very happy to hear from you.

I have been fortunate to meet many friends and mentors who have helped me along the way. I am also always happy to chat and offer whatever help I can to others.

If you have questions about my research, or if you have tried to reach me through GitHub issues but have not received a response, please feel free to email me directly.

My preferred email: niuyuwei04@gmail.com