Yang Zhou (周洋)

I am a 2nd-year Ph.D. student in the College of Computer Science at Fudan University, advised by Dr. Tong He. I received my Bachelor of Computer Science and Technology from Sichuan University in June 2024.

My research interests lie at the intersection of world modeling, reinforcement learning, and data-centric AI. I am broadly interested in building 4D world models for perception, reconstruction, prediction, and planning, supported by scalable multimodal data infrastructure. More recently, I have also been exploring RL-based alignment for diffusion models and large language models.

Feel free to drop an email to reach out! (discussion, collaboration, etc.)

Email / Scholar / GitHub / X

Publications

OmniWorld: A Multi-Domain and Multi-Modal Dataset for 4D World Modeling

Yang Zhou, Yifan Wang, Jianjun Zhou, Wenzheng Chang, Haoyu Guo, Zizun Li, Kaijing Ma, Xinyue Li, Yating Wang, et al.

ICLR, 2026

paper / project page / code / dataset

A multi-domain and multimodal dataset designed to support large-scale 4D world modeling across robotics, simulation, human activities, and in-the-wild scenarios.

π3: Permutation-Equivariant Visual Geometry Learning

Yifan Wang*, Jianjun Zhou*, Haoyi Zhu, Wenzheng Chang, Yang Zhou, Zizun Li, Junyi Chen, Jiangmiao Pang, Chunhua Shen, Tong He

ICLR, 2026

paper / project page / code

A geometry learning framework with permutation-equivariant design for visual reasoning and structured world understanding.

Wint3r: Window-based Streaming Reconstruction with Camera Token Pool

Zizun Li, Jianjun Zhou, Yifan Wang, Haoyu Guo, Wenzheng Chang, Yang Zhou, Haoyi Zhu, Junyi Chen, Chunhua Shen, Tong He

ICLR, 2026

paper / project page / code

A streaming reconstruction method that improves long-horizon scene modeling with a window-based design and camera token pooling.

Aether: Geometric-Aware Unified World Modeling

Haoyi Zhu*, Yifan Wang*, Jianjun Zhou*, Wenzheng Chang*, Yang Zhou*, Zizun Li*, Junyi Chen*, Chunhua Shen, Jiangmiao Pang, Tong He

ICCV, 2025. Best Paper Award, RIWM workshop.

paper / project page / code

A unified world modeling framework that incorporates geometry-aware representations for perception, prediction, and generation.

DeepVerse: 4D Autoregressive Video Generation as a World Model

Junyi Chen, Haoyi Zhu, Xianglong He, Yifan Wang, Jianjun Zhou, Wenzheng Chang, Yang Zhou, Zizun Li, Zhoujie Fu, Jiangmiao Pang, Tong He

arXiv, 2025

paper / project page / code

An autoregressive video generation framework that treats 4D video synthesis as a world modeling problem.

HiSplat
HiSplat: Hierarchical 3D Gaussian Splatting for Generalizable Sparse-View Reconstruction

Shengji Tang, Weicai Ye, Peng Ye, Weihao Lin, Yang Zhou, Tao Chen, Wanli Ouyang,

ICLR, 2025

paper / project page / code

A hierarchical Gaussian splatting approach for robust and generalizable sparse-view 3D reconstruction.

Projects

RealGRPO
RealGRPO: A Simple Way to Eliminate Reward Hacking in GRPO Diffusion Alignment

Yang Zhou, Haoyu Guo

Feb. 2026

code

RealGRPO addresses reward hacking in GRPO-based diffusion alignment by using an LLM to dynamically generate contrastive style pairs.

Experience

Shanghai Artificial Intelligence Laboratory

Research Intern

Supervisors: Tong He · Wanli Ouyang

Shanghai, China
Dec. 2023 -- Present

  • Worked on 4D world models for perception, reconstruction, prediction, and planning.
  • Contributed to large-scale multimodal data infrastructure and annotation pipelines for world modeling.
  • Explored RL-based alignment for diffusion models and large language models.

Professional Services

Thanks Jon Barron for sharing the source code of this website template.