Yunfeng Wu

Results 1 issues of Yunfeng Wu

How can we design a memory-friendly mask of the linear attention implemented in Sana-Video?