nerf
nerf copied to clipboard
MLP documentation
Why MLP uses fully-connected blocks instead of convolution blocks? What are the advantages of fully-connected blocks over convolution blocks?Could you elaborate on that.