efficient-attention
efficient-attention copied to clipboard
cmsflash
An implementation of the efficient attention module.
Can you share the code on how to visualize attention maps in object detection like the one shown in your paper? 
the input dimension is (1, 128, 32, 32) or (1, 256, 16, 16). I try to use efficient attention ,but the memory and Gflops are still big
Hello, I have recently implemented a cross attention application with multi-modal fusion, but because the image resolution is too large, cuda OOM occurs when calculating q and k, so I...
