About
Hi! 👋 I am Yihang Du (杜艺航). I am an incoming PhD student at FreedomAI Lab@CUHK-Shenzhen, advised by Prof. Benyou Wang. I also work with Dr.Yan Hu.
My research focuses on efficient and generalizable generative models, with a particular interest in representation learning across different model architectures (e.g., autoregressive, diffusion-based, unified multimodal, and native multimodal architectures). I aim to understand how these models acquire, organize, and utilize internal representations, how such representations support generation and generalization, and how insights into their representational mechanisms can guide the development of more capable and efficient foundation models.
I am actively seeking internship and research collaboration opportunities in MLLMs and agentic AI. Please feel free to reach out if you are interested in collaborating.
Update
Older updates Hide older updates
Publications
[Google Scholar] († denotes equal contribution)
-
Why Vision Fails as a Universal Bridge: Rectifying Modality Asynchrony in Multilingual MLLMsCOLM 2026
-
Nature Communications 2024
Education
Experience
Misc
I am a big fan of Podcasts and I enjoy playing table tennis. I am trying to learn tennis.
I love photography. I capture the beauty of my travels with a SONY A7M4. Here are some of my photos.