柯磊
First-year Ph.D. Student
Hong Kong University of Science and Technology
Hi, I'm Lei Ke, a first-year Ph.D. student at The Hong Kong University of Science and Technology (HKUST), supervised by Prof. Fangneng Zhan and Prof. Qifeng Chen.
Prior to that, I obtained my M.Sc. degree (2023–2026) from Tsinghua University, supervised by Professor Yujiu Yang. I received my B.Eng. degree (2019–2023) from Shanghai Jiao Tong University.
I am working on generative models, with a focus on world models, video generation, and diffusion acceleration.
Publications
Real-time interactive audio-visual world generation via causal distillation.
Full-song music generation that composes a readable score before rendering audio.
Empower the unified anything-to-audio generation with step distillation.
Embodied egocentric world simulation with view-based evolution customization.
Diffusion-native latent reward modeling for alignment without VLMs.
Bring ReFlow-based step distillation to SOTA.
Advanced diffusion acceleration method on Stable Diffusion.
* Equal contribution. † Corresponding author.
Education
Sept. 2026 – Present
First-year Ph.D. Student · Hong Kong 香港科技大学
Supervised by Fangneng Zhan and Qifeng Chen
Sept. 2019 – Jun. 2023
B.Eng. · Shanghai
Internships
Feb. 2026 – Present
Research Intern
Jun. 2025 – Dec. 2025
Algorithm Intern, WeChat BG – Technical Architecture Dept.
Jan. 2025 – May 2025
Algorithm Intern, Large Model & Multimedia Tech Dept.