Guande He (何冠德)

photo_me.jpg

AI Researcher, ??????😎

I am a Founding Member of a stealth startup. Before that, I was a Ph.D. student at the McCombs School of Business, The University of Texas at Austin, working with Prof. Mingyuan Zhou. I was a Master’s student at the TSAIL Group, Department of Computer Science and Technology, Tsinghua University, where I was honored to be advised by Prof. Jun Zhu and Prof. Jianfei Chen. I received my B.Eng. from the School of Software, Tsinghua University in 2021. In summer 2025, I was a research intern at Adobe Research under Dr. Eli Shechtman’s team.

My research interests include machine learning and foundation models. I am currently developing principled post-training algorithms (e.g., distillation, alignment, and self-play refinement) and efficient sampling techniques for deep generative models.

Selected Publications & Preprints [full list]

* denotes equal contribution; † denotes corresponding authors

  1. arXiv
    Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion
    Xun Huang , Zhengqi Li,  Guande He , Mingyuan Zhou, and Eli Shechtman
    arXiv preprint, arXiv:2506.08009, 2025
    PDF Code Website
  2. ICML
    RIFLEx: A Free Lunch for Length Extrapolation in Video Diffusion Transformers
    Min Zhao,  Guande He , Yixiao Chen , Hongzhou Zhu, Chongxuan Li, and Jun Zhu
    In Forty-Second International Conference on Machine Learning, Vancouver, Canada, 2025
    PDF Code Website
  3. ICML
    Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator
    Kaiwen Zheng , Yongxin Chen , Huayu Chen,  Guande He , Ming-Yu Liu, Jun Zhu, and Qinsheng Zhang
    In Forty-Second International Conference on Machine Learning, Vancouver, Canada, 2025

    Spotlight (Accept rate 2.6%)

    PDF Website
  4. ICLR
    Diffusion Bridge Implicit Models
    Kaiwen Zheng*,  Guande He* , Jianfei Chen, Fan Bao, and Jun Zhu†
    In The Thirteenth International Conference on Learning Representations, Singapore, 2025
    PDF Code
  5. arXiv
    Vidu: a Highly Consistent, Dynamic and Skilled Text-to-Video Generator with Diffusion Models
    Fan Bao, Chendong Xiang*, Gang Yue*,  Guande He* , Hongzhou Zhu*, Kaiwen Zheng*, Min Zhao*, Shilong Liu*, Yaole Wang*, and Jun Zhu†
    Technical Report, arXiv:2405.04233, 2024
    PDF Website
  6. NeurIPS
    Consistency Diffusion Bridge Models
    Guande He* , Kaiwen Zheng*, Jianfei Chen, Fan Bao, and Jun Zhu†
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems, Vancouver, Canada, 2024
    PDF
  7. NeurIPS
    Noise Contrastive Alignment of Language Models with Explicit Rewards
    Huayu Chen,  Guande He , Lifan Yuan , Ganqu Cui, Hang Su, and Jun Zhu†
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems, Vancouver, Canada, 2024
    PDF Code
  8. ICLR
    Preserving Pre-trained Features Helps Calibrate Fine-tuned Language Models
    Guande He , Jianfei Chen†, and Jun Zhu†
    In The Eleventh International Conference on Learning Representations, Kigali, Rwanda, 2023
    PDF Code

Experience

Shengshu Technology

Research Intern

2023.11 - 2024.07

Beijing, China

Teaching

  • 2024 Fall, TA in STA 235 "Data Science for Business Applications", UT Austin, instructed by Prof. Mingyuan Zhou.
  • 2023 Spring, TA in "Statistical Learning Theory and Applications", Tsinghua University, instructed by Prof. Jun Zhu.

Miscellaneous

My slides for TSAIL reading group:

Other slides I made for presenting ML papers:

Personal

I am an amateur saxophone and (bass) trombone player and I enjoy playing music 🎶 :-)