About me

I am Xinyao Liao, a third-year master’s student in Computer Science at Huazhong University of Science & Technology (HUST), supervised by Prof. Wei Wei. I am currently a research intern in the CVML Lab at the National University of Singapore (NUS), advised by Prof. Angela Yao. Previously, I completed a B.E. in Information Security and a B.A. in Philosophy, both at HUST.

My recent research focuses on visual generative models and reinforcement learning. I am also interested in broader questions related to Artificial General Intelligence (AGI), including the design of general training objectives and learning paradigms.

News

[Feb 2026] VA-π accepted to CVPR 2026!
[Jan 2026] reAR accepted to ICLR 2026!
[Jul 2025] Joined CVML Lab at NUS as a research intern, advised by Prof. Angela Yao.
[Jul 2024] UniQ accepted to ACM MM 2024!
[Sep 2023] Started my master’s degree at HUST.

Publications

VA-π: Variational Policy Alignment for Pixel-Aware Autoregressive Generation

CVPR 2026

Website / Paper / Code

Xinyao Liao* , Qiyuan He* , Kai Xu , Xiaoye Qu , Yicong Li , Wei Wei , Angela Yao

reAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization

ICLR 2026

Website / Paper / Code

Qiyuan He , Yicong Li , Haotian Ye , Jinghao Wang , Xinyao Liao , Pheng-Ann Heng , Stefano Ermon , James Zou , Angela Yao

CoCA: Step-Level Reward for Free in RL-Based T2I Diffusion Fine-Tuning

Arxiv 2025

Paper / Code

Xinyao Liao , Wei Wei , Xiaoye Qu , Qiyuan He , Angela Yao , Yu Cheng

UniQ: Unified Decoder with Task-specific Queries for Efficient Scene Graph Generation

ACM MM 2024

Paper / Code

Xinyao Liao , Wei Wei , Dangyang Chen , Yuanyuan Fu

Other than research, let’s read, watch movies, listen to music, and create fun things!

Xinyao Liao