MaskCLIP
MaskCLIP copied to clipboard
About 1x1 convolution
Why does MakClip's 1x1 convolution need no training? Also, why can't I see the 1x1 position in the code?