I found a layer (separable convolution) implemented in CuPy. This layer was developed as a part of a pytorch project. Is it possible to use the same CuPy code as a layer in a tensorflow model?
I am looking for a generic solution that can be applied to any custom layer implemented in CuPy.
question from:https://stackoverflow.com/questions/65948623/how-to-integrate-custom-layer-implemented-with-cupy-in-a-tensorflow-model