Feature request: traspose convolution, 1D convolutions, causal convolutions

[pinouchon]

In deep learning setup I am experimenting with fancy convolutions, such as transpose conv, 1D and causal. (For an example of causal conv, see this: https://github.com/locuslab/TCN/blob/ec2ee1038c941a2f56af66915135a1a892043aed/TCN/tcn.py#L6, you shift the input so that it is "centered right" instead of "centered middle"). I would love to have these available for easier experimentation.

But I see that in your readme you already have a long list of features you would like to add, so I may have do write those features myself...

[ezyang]

I'll be happy to merge PRs :) Or if you are inspired by this to build your own version, even better! Note that transposed convolution, you just read the diagram "backwards" (input is output, and vice versa)

[whubaichuan]

@pinouchon @ezyang Hi, will the project add the 3D convolution-visualizer?

[ezyang]

Probably never, it would be a pretty major rewrite since I'd have to design some sort of 3d format