Shu Pu

I am a first year Ph.D. student of VLAA Lab at University of California, Santa Cruz, advised by Professor Cihang Xie and Yuyin Zhou. Before that, I received my bachelor degree at Huazhong University of Science and Technology (HUST) in Wuhan, where I had the privilege of working with Professor Yao Wan. I was honored to work with Professor Zhuang Liu as a research intern. I'm interested in representation learning and generative AI, with a particular interest in world models.

Email  /  CV  /  Bio  /  Scholar  /  Github  /  Transcript  /  Blog

profile photo

Selected Research

Memorization in 3D Shape Generation: An Empirical Study
Shu Pu, Boya Zeng, Kaichen Zhou, Mengyu Wang, Zhuang Liu
CVPR Findings, 2026
ArXiv / Code

An empirical study of memorization in 3D shape generation models.

Judge Anything: MLLM as a Judge Across Any Modality
🎉🎉🎉 Accepted by ACM KDD 25' DB Track Oral!
Shu Pu, Yaochen Wang, Dongping Chen, Zhiyuan Zhang*, Zetong Zhou*, Shuang Gong*, Yuhang Chen*, Qi Qin*, Zhongyi Zhang*, Guohao Wang*, Yi Gui, Yao Wan, Philip S. Yu
KDD, 2025
ArXiv / Code

A system analysis for utilizing MLLM as a judge across any modality and an Omni-Modality competitive arena.

ISGBench: Interleaved Scene Graph for Interleaved Text-and-Image Generation Assessment
🎉🎉🎉 Accepted by ICLR 25' Spotlight!
Dongping Chen*, Ruoxi Chen*, Shu Pu*, Zhaoyi Liu*, Yanru Wu*, Caixi Chen*, Benlin Liu, Yue Huang, Yao Wan, Pan Zhou, Ranjay Krishna
ICLR, 2025
ArXiv / Code

A comprehensive evaluation framework for interleaved text-and-image generation.


Thanks to this Jon Barron's website template to create such a clean and neat personal website. The source code is open-source. Do not scrape the HTML from this page itself, as it includes analytics tags that you do not want on your own website — use the github code instead. Also, consider using Leonid Keselman's Jekyll fork of this page.