Xiangyue Zhang (章湘粤) | Embodied AI, Motion Generation

News

Nov 1, 2026

🚀 NextTransformer is now open source!

Sep 24, 2026

🎉 PASPA was accepted by NeurIPS 2026.

Jun 18, 2026

🎉 StreamTalk, OmniDance, and FlowerDance were accepted by ECCV 2026.

Mar 31, 2026

🎉 MACE-Dance was accepted by SIGGRAPH 2026.

Nov 8, 2025

🎉 GlobalDiff was accepted by AAAI 2026.

Jun 26, 2025

🎉 SemTalk was accepted by ICCV 2025.

Dec 22, 2024

🥂 Our band performed successfully at the Hua Young Music Festival! Cheers! Check More to see pictures.

Open-Source Systems

Open Source · 2026

NextTransformer: An Open Framework and Fair Benchmark for Next-Generation Transformer Architectures

A shared training framework and compute-matched benchmark for comparing Transformer architectures, with an open results observatory.

Code Website Docs
RSI Project 2026

Deep Researcher Agent: An Autonomous Framework for 24/7 Deep Learning Experimentation with Zero-Cost Monitoring

Xiangyue Zhang

arXiv preprint arXiv:2604.05854, 2026

An autonomous research framework for round-the-clock deep learning experiments, from implementation and monitoring to analysis and refinement.

arXiv Website Code Architecture Guide

We present Deep Researcher Agent, an open-source framework that enables large language model agents to autonomously conduct deep learning experiments around the clock, spanning the full lifecycle from hypothesis formation and code implementation to result analysis and iterative refinement.

The paper introduces zero-cost monitoring during training, a two-tier constant-size memory capped at roughly 5K characters, and a minimal-toolset leader-worker architecture for lower token overhead in long-running experiment automation.

Preprints

Preprint · 2026

World2Motion: Turning Video World Models into 3D Human Motion Generators

Fangyuan Tu, Xiangyue Zhang, Yiyi Cai, Yichen Peng, Kunhang Li, Bo Zheng, Zhixiang Wang, Kaipeng Zhang, Erwin Wu, Haoran Xie, Haiyang Liu

arXiv preprint, 2026

arXiv Bib Website

@article{tu2026world2motion,
  title={World2Motion: Turning Video World Models into 3D Human Motion Generators},
  author={Tu, Fangyuan and Zhang, Xiangyue and Cai, Yiyi and Peng, Yichen and Li, Kunhang and Zheng, Bo and Wang, Zhixiang and Zhang, Kaipeng and Wu, Erwin and Xie, Haoran and Liu, Haiyang},
  journal={arXiv preprint arXiv:2609.37004},
  year={2026},
  url={https://arxiv.org/abs/2609.37004}
}

Preprint · 2026

FloodDiffusion 2: Efficient and Path Controllable Streaming Motion Generation

Yiyi Cai, Yuhan Wu, Kunhang Li, Tu Fangyuan, Xiangyue Zhang, Qiaoge Li, Zhixiang Wang, Kaipeng Zhang, Haiyang Liu

arXiv preprint, 2026

arXiv Bib Code Website

@article{cai2026flooddiffusion2,
  title={FloodDiffusion 2: Efficient and Path Controllable Streaming Motion Generation},
  author={Cai, Yiyi and Wu, Yuhan and Li, Kunhang and Tu, Fangyuan and Zhang, Xiangyue and Li, Qiaoge and Wang, Zhixiang and Zhang, Kaipeng and Liu, Haiyang},
  journal={arXiv preprint arXiv:2609.33167},
  year={2026},
  url={https://arxiv.org/abs/2609.33167}
}

Selected Publications

🔥 ECCV 2026

StreamTalk: Streaming Co-Speech Gesture Generation with Key-Pose Anchoring

Xiangyue Zhang, Jianfang Li, Jiaxu Zhang, Kaixing Yang, and Steven Hoi

European Conference on Computer Vision (ECCV), 2026

arXiv Bib Code Website Video


@inproceedings{zhang2026streamtalk,
  title={StreamTalk: Streaming Co-Speech Gesture Generation with Key-Pose Anchoring},
  author={Zhang, Xiangyue and Li, Jianfang and Zhang, Jiaxu and Yang, Kaixing and Hoi, Steven},
  booktitle={European Conference on Computer Vision},
  year={2026},
  eprint={2608.01643},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2608.01643}
}

🔥 AAAI 2026

Mitigating Error Accumulation in Co-Speech Motion Generation via Global Rotation Diffusion and Multi-Level Constraints

Xiangyue Zhang, Jianfang Li, Jianqiang Ren, and Jiaxu Zhang

Annual AAAI Conference on Artificial Intelligence (AAAI), 2026

arXiv Bib Code Website Video Results


@misc{zhang2025mitigatingerroraccumulationcospeech,
      title={Mitigating Error Accumulation in Co-Speech Motion Generation via Global Rotation Diffusion and Multi-Level Constraints}, 
      author={Xiangyue Zhang and Jianfang Li and Jianqiang Ren and Jiaxu Zhang},
      year={2025},
      eprint={2511.10076},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2511.10076}, 
}

🔥 ICCV 2025

SemTalk: Holistic Co-speech Motion Generation with Frame-level Semantic Emphasis

Xiangyue Zhang, Jianfang Li, Jiaxu Zhang, Ziqiang Dang, Jianqiang Ren, Liefeng Bo, and Zhigang Tu

International Conference on Computer Vision (ICCV), 2025

arXiv Bib Code Website Video Results


@inproceedings{zhang2025semtalk,
  title={SemTalk: Holistic Co-speech Motion Generation with Frame-level Semantic Emphasis},
  author={Zhang, Xiangyue and Li, Jianfang and Zhang, Jiaxu and Dang, Ziqiang and Ren, Jianqiang and Bo, Liefeng and Tu, Zhigang},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={13761--13771},
  year={2025}
}

Experience & Education

2026.10 - PresentTokyo

The University of Tokyo

Ph.D. Student, Mechano-Informatics, IST

Advisor: Prof. Tatsuya Harada.

2026.06 - PresentShenzhen

Tencent

Research Intern, Qingyun Program (青云)

Working on world-motion models and agentic AI for games.

2025.12 - 2026.03Shenzhen

ByteDance

Research Intern, Intelligent Creation Team

Working on large streaming motion generation models.

2024.06 - 2025.11Hangzhou

Alibaba

Research Intern, Tongyi Lab

Advisors: Dr. Liefeng Bo and Dr. Jianfang Li. Working on co-speech motion generation.

2023.09 - 2026.06Wuhan

Wuhan University

Master Student, Computer Application

Advisor: Prof. Zhigang Tu.

2019.09 - 2023.06Changsha

Central South University

B.S. Degree

Minored in Data Science and Big Data Technology.

Awards & Honors

2026 - 2029

The University of Tokyo Fellowship (top 1%)

JUN, 2026

Outstanding Graduate Award (Master's)

JAN, 2026

Wang Zhizhuo Scholarship for Innovative Talents (top 0.3%)

OCT, 2025

National Scholarship (top 3%)

JUN, 2023

Outstanding Graduate Award (Bachelor's)

Service

Reviewer

Reviewer Service: ICLR, NeurIPS, AAAI, ACM MM, T-CSVT, PR