Have `Conv1d`, `Conv2d`, `Conv3d` instead of `Conv`? Same for `Pool`?

[albertz]

It's again a question on explicite vs implicit. Currently the kernel size defines the dimension. But this is maybe easier to read in the code, so that it is exactly clear what type of convolution you do.

PyTorch has Conv1d, Conv2d, Conv3d.

Flax has just Conv.

[albertz]

It was done now for Conv, i.e. we have Conv1d, Conv2d, Conv3d, and same for TransposedConv, i.e. TransposedConv1d, TransposedConv2d, TransposedConv3d, via a2e382c.

[albertz]

pool is not a module but just a function. The same question remains though.

[albertz]

Ok, now we have pool1d, pool2d, pool3d.

I guess this issue can be closed.