Open
Description
I applied the MambaVisionBlock to object detection,the training speed is particularly slow,about 4.3s/it,if i use yolo or rt-detr,the speed is about 2it/s,The difference in speed is 8 times.But the number of parameters is about the same,about 35% more computation.Have you ever been in the same situation?
Metadata
Metadata
Assignees
Labels
No labels