More encoder/decoder examples in the 'quickstart'

[jarvist]

Currently the one example shows building a Convolutional auto-encoder, and training it on the MNIST dataset.

I think it would be really useful if there were some other examples of using the package. For instance on character based data, or graphs. I think for any application, the first thing a user of the package would do is look to see how they could build an encoder/decoder for the structure of their data.

[mrazomej]

That's a totally reasonable suggestion, @jarvist. Do you have a particular example you like when thinking about VAEs? If you point me to a dataset or a specific application, I'll try to implement it with AutoEncoderToolkit. Or even better, if you want to do a PR with your favorite example, that would be greatly appreciated.

[jarvist]

For me the obvious one is to use text. There is a Python/PyTorch resource based on a 2019 paper which I generally direct students towards: https://github.com/shentianxiao/text-autoencoders/tree/master

Certainly, if myself or my students make anything, we'll be sure to create an example out of it & make a pull request!