A problem found in the code
Closed this issue · 1 comments
eezywu commented
Should nn.init.constant(self.W[1].weight, 0) be changed to nn.init.constant(self.W[0].weight, 0)?
AlexHex7 commented
I want to set the parameters γ and β in Batch Normalization to 0, so that the output of it will be zero. Then the output of the Non-local block is equal to the input. In this way, this block can be inserted to any models without change the behavior at the first iteration.