Zian Li (李自安)

Zian Li (李自安)

Done is better than perfect.

🧑‍💻 Biography

I am now a fourth-year Ph.D. student at the Institute for Artificial Intelligence, Peking University, advised by Prof. Muhan Zhang. I have interned at the Kling Team and the HappyHorse Team for video generation research and indursrial model development.

🔬 Research Interests

My research focuses on scalable generative models spanning multiple modalities, including videos, images, and molecules. I am interested in both theoretical foundations (e.g., representations expressiveness) and practical systems (e.g., architectures, acceleration, and industrial deployment). Recently, I have been particularly interested in video generation. My research interests include:

  • Video Generation: architectures, pretraining paradigms, few-step distillation, refiners, and generation-friendly VAEs.
  • AI for Science: geometric representations, molecule pretraining and generation paradigms, few-step generation, ai for biology.

If you are interested in these areas, feel free to reach out!

🎖 Honors and Awards

2025.10, Doctoral Dean’s Scholarship, Institute for Artificial Intelligence, Peking University
2020.12, 2021.12, 2022.12, National Scholarship for Undergraduate Student
2021.12, Outstanding Student Pioneer of Tianjin University (Nomination Award) (only 10 awardees and 5 nominated awardees university-wide per year)
2021.08, First Prize of National Zhou Peiyuan Competition on Mechanics (~0.3%)
2021.04, Meritorious Winner (First Prize) of ICM: Interdisciplinary Contest In Modeling

📖 Educations

💻 Internships

📃 Publications

* indicates equal contribution.
preprint
DUET: A Diversity-Quality Duet of Distillation Experts for Two-Step Video Generation
Zian Li, Litong Gong, Borui Liao, Pengfei Liu, Xinyu Wang, Xinyuan Wei, Yifan Gao, Tiezheng Ge, Muhan Zhang
NeurIPS 2026
CanvasMAR: Improving Masked Autoregressive Video Generation With Canvas
Zian Li, Muhan Zhang
NeurIPS 2026
Toward Better Geometric Representations for Molecule Generative Models
Shaoheng Yan*, Zian Li*, Cai Zhou, Qiaojing Huang, Kai Liu, Muhan Zhang
NeurIPS 2026
FlashMol: High-Quality Molecule Generation in as Few as Four Steps
Xinyuan Wei*, Zian Li*, Shaoheng Yan, Cai Zhou, Muhan Zhang
preprint
In-Parameter Learning: Why Lifelong AI Systems Need More Than Longer Context
Yiding Wang*, Haotong Yang*, Pingzhi Tang*, ..., Zian Li, ..., Xing Sun†, Muhan Zhang†‡
ACMMM 2026
SubjectAnchor: Subject-Aware Memory-to-Video for Multi-Shot Storytelling
Xinyu Wang, Huafeng Shi, Zian Li, Yan Zhou, Xiaoqiang Liu, Yue Ma, Pengfei Wan
preprint
Rethinking Diffusion Models with Symmetries through Canonicalization with Applications to Molecular Graph Generation
Cai Zhou, Zijie Chen, Zian Li, Jike Wang, Kaiyi Jiang, Pan Li, Rose Yu, Muhan Zhang, Stephen Bates, Tommi Jaakkola
ICML 2026 GenBio Workshop Spotlight
GeoRecon: Graph-Level Representation Learning for 3D Molecules via Reconstruction-Based Pretraining
Shaoheng Yan, Zian Li, Muhan Zhang
ICML 2025 Spotlight
Geometric Representation Condition Improves Equivariant Molecule Generation
Zian Li*, Cai Zhou*, Xiyuan Wang, Xingang Peng, Muhan Zhang
ICLR 2025
On the Completeness of Invariant Geometric Deep Learning Models
Zian Li, Xiyuan Wang, Shijia Kang, Muhan Zhang
NeurIPS 2023
Is Distance Matrix Enough for Geometric Deep Learning?
Zian Li, Xiyuan Wang, Yinan Huang, Muhan Zhang