Yijun Liu

Ultra Flash achieves real-time high-resolution video generation at unprecedented frame rates, pushing the boundaries of what’s possible in streaming video AI.

LLMs may outshine human instructors in feedback ratings, but their complexity masks critical differences in targeted sentence feedback.

Unlock the full potential of your pretrained video diffusion models with a surprisingly simple four-stage post-training framework that drastically improves visual quality, temporal coherence, and instruction following.