How to Multiply 2 feature maps from 2 different cnn layers


I am trying to get a feature map from VGG11 conv5 layer as (14 X 14 X 512) and also getting another feature map from another different layer as (1 X 1 X 512). I want to multiply these maps and get the output as (14 X 14 X 512) and pass it to FC layer. Is there are any layer that can do multiplication like this?

just torch.mul(x1, x2) will do, where x1, x2 are those two features.

Does it support backprop?

yes it supports backprop.