Shikang Zheng, Jingkai Huang, Jiacheng Liu +5
Diffusion models have achieved impressive performance in video generation, but their iterative denoising process remains computationally expensive due to the large number of tokens…