Adapt to changes in NNlib to use mode externally

[ayush1999]

This works on the changes made in FluxML/NNlib.jl#53 .
(To use mode directly from Flux).

[MikeInnes]

I think the API needs some thought here, and I'd rather not 0/1 (which are unclear to read). Maybe a kwarg like reflect=true?

I'm almost tempted to just have a separate CrossCorrelation layer for this, but it's probably not worth that.

[ayush1999]

@MikeInnes I've added reflect as an attribute of Conv. When true, reflects internally acts similar to mode=0.

[philtomson]

It seems like convolution mode should be the default with cross correlation as the option.

[MikeInnes]

I believe we already do; mode = 0 corresponds to convolution in CUDNN.

[ayush1999]

@philtomson The nomenclature is actually a bit confusing. Convolution means flipping the kernel and then adding the element wise products. Cross Correlation is the same as Convolution, but without flipping the kernel. Flux by default flips the kernel, and hence does Convolutions. I think it'd be better to have a kwarg with mode="convolution" or mode="cross-correlation", so that there isn't any confusion.

[philtomson]

Yeah, I don't like plain strings for these kinds of things (It's all the rage in Python-land I know, but I think we can do better).

It would be good if it was some kind of enum type, I think you can do this with @enum:

@enum Choice convolution correlation

EDIT: actually, thinking about it a bit more, I think I like the idea of a separate CrossCorrelation layer/function for this. It would just be a wrapper around Conv with the kernel flipped and passed into Conv. It just makes it cleaner all around to make it explicit, I think, and no special-casing needed in the the actual convolution function which will get used more often anyway.

[MikeInnes]

Are we doing #423 instead?

[ayush1999]

Yes, there was a plan to add an option to Conv initially, but that changed later. Closing this.