For a convolutional kernel is of size ©x(H)x(W), I want all the channels within a same (h,w) position to share a same dropout mask. How can i manage it using the dropout function?
1 Like
For a convolutional kernel is of size ©x(H)x(W), I want all the channels within a same (h,w) position to share a same dropout mask. How can i manage it using the dropout function?