This PR add a Concatenate layer equivalent to keras.layers.concatenate to concat the outputs of a list of layers. This type of layers is useful for architectures of type inception where different kernel sized convs are stacked together. I inherited from nn.ModueList but it could also inherit from fastai Module.
This PR add a Concatenate layer equivalent to
keras.layers.concatenate
to concat the outputs of a list of layers. This type of layers is useful for architectures of type inception where different kernel sized convs are stacked together. I inherited fromnn.ModueList
but it could also inherit from fastaiModule
.