1 paper
Xingyu Jia, Baole Ai, Ang Wang +2
Diffusion Transformers (DiTs) enable high-quality video generation but suffer from substantial inference latency, primarily attributable to the computationally expensive full spati…