Non-descriptive arg in Conv: why filter intead of size?

[bhvieira]

One of Conv parameters is now called filter (after #873 was merged.
In my opinion, size better described what Flux expects to be supplied by the user: the size of the kernel. Now it reads filter, which, to me, sounds like Flux expects the user to supply the actual kernel here, not its size.
If others think it's fine, then I'm ok with that too.

[levimcclenny]

Curious where you see this? The source and Flux.jl docs say size from what I'm seeing, unless you're seeing this somewhere else:

https://fluxml.ai/Flux.jl/stable/models/layers/#Flux.Conv
https://github.com/FluxML/Flux.jl/blob/7a32a703f0f2842dda73d4454aff5990ade365d5/src/layers/conv.jl

To address your question specifically, Keras uses kernel_size which seems really clear to me. size would probably be preferred to filter if describing the size of convolutional filters

[DhairyaLGandhi]

filter_size seemed a bit long, but essentially is the intended reading of it. Happy to change it to a nicer / better descriptive name, of course.

@levimcclenny you can go to the drop-down on the bottom of the table of contents and select dev to get the dev docs of the unlreleased version.

[levimcclenny]

Ah I see now, looking at the example I think it's reasonably clear what the argument is, but without the example could be somewhat confusing. Personal opinion, obviously.