Carlo Lucibello
Carlo Lucibello
I'm not sure I understand. Can you make an example or define more precisely the expected behavior?
But MLDatasets contains generic datasets, while Metalhead explicitly targets vision. Since vision models are commonly used with preprocessing, I think it would make more sense to have some preprocessing pipeline...
Maybe we should have a VisionDatasets.jl package then
> Is the concern here about MLDatasets deps? yes > It already provides the common vision datasets except ImageNet, so I think I'm missing why vision should be separate. If...
It's generally useful. Pytorch and tensorflow have it, deep graph library and pytorch geometric use it internally and I need it for https://github.com/CarloLucibello/GraphNeuralNetworks.jl. I'll look into and eventually port here...
I can't find topk in Metalhead.jl, nor in ResNetImageNet.jl
I consider it a differentiable transformation, a generalization of maximum, and prefer having it here for wider availability. We also have `softmax` here, which is a bit related as well
We are likely removing the forced conversion in #212. That said I don't think NNPACK has Complex support, right? We could support complex convolution in `conv_im2col` I guess
the inconsistency has been fixed in #238, leaving this open for the missing pooling optinos
from a quick glance looks mostly ok, thanks. Convenience methods for the 1d and 2d cases are missing. Then some tests and `rrule`s