Independent project. Not a U.S. government website.

Support our project (opens Buy Me a Coffee)
USASI

Peiyuan Zhang

First author of the Sliding Tile Attention and VSA papers cited by FastVideo; Hao AI Lab alumnus

  • Works at ByteDance (title not stated), ByteDance45

Peiyuan Zhang is the first author of two video-generation papers from UC San Diego's Hao AI Lab, Sliding Tile Attention and VSA (Video Sparse Attention), which the FastVideo README asks users to cite. Zhang was a PhD student in the lab who left with a master's degree, and the lab's alumni list shows Zhang now at ByteDance.1234

Research

With Hao Zhang and co-authors, Zhang posted "Fast Video Generation with Sliding Tile Attention" in February 2025 and "VSA: Faster Video Diffusion with Trainable Sparse Attention" in May 2025, both as first author. Both methods target the cost of attention in video diffusion models.12

The FastVideo README records the release of Sliding Tile Attention inference code in February 2025 and VSA fine-tuning and inference code in June 2025, and asks users of the framework to cite both papers.3

Career

Zhang was a PhD student in the Hao AI Lab at UC San Diego and left the program with a master's degree. The lab's people page and Hao Zhang's site list Zhang among alumni now at ByteDance.45

Work

  • FastVideoOpen-source code

    First author of the Sliding Tile Attention and VSA papers, the two research works the FastVideo README lists for citation; the README notes released inference code for Sliding Tile Attention and fine-tuning and inference code for VSA.312

    Project page (external site: github.com)

Career timeline

  1. Feb 2025

    Posted "Fast Video Generation with Sliding Tile Attention" as first author.1

  2. May 2025

    Posted "VSA, Faster Video Diffusion with Trainable Sparse Attention" as first author.2

  3. Jun 14, 2025

    FastVideo released fine-tuning and inference code for VSA.3

  4. 2026

    Listed by the Hao AI Lab as an alumnus now at ByteDance.5

Other documented roles

Roles the sources document but that are not confirmed as current.

More sourced profiles at the same organization, in the same field, or from the same school.

Is this you?

Profiles cover public professional work only, and every line cites a source. If something is out of date, missing or wrong, send a correction with a link to an official page and we will update it. Suggest a correction

Profile on USASIAdd this badge to your website or GitHub profile README. It links back to this page. It is not an endorsement or a rating.

Markdown (for a README)

[![Peiyuan Zhang — Profile on USASI](https://unitedstatesofamericasuperintelligence.com/badges/profile.svg)](https://unitedstatesofamericasuperintelligence.com/people/peiyuan-zhang/)

HTML

<a href="https://unitedstatesofamericasuperintelligence.com/people/peiyuan-zhang/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/profile.svg" alt="Peiyuan Zhang — Profile on USASI" height="20"></a>

This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.

References · reviewed Oct 22, 2026

  1. 1.
    Fast Video Generation with Sliding Tile Attention (arXiv 2502.04507) (external site: arxiv.org)

    arXiv · Paper · published Feb 6, 2025 · accessed Oct 22, 2026

  2. 2.
  3. 3.
    hao-ai-lab/FastVideo README (news and citation) (external site: github.com)

    Hao AI Lab (GitHub repository) · Repository · accessed Oct 22, 2026

  4. 4.
    Hao AI Lab — People (alumni) (external site: haoailab.com)

    Hao AI Lab, UC San Diego · Official page · accessed Oct 22, 2026

  5. 5.

Support Us

Help keep USASI useful.

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project