VQVAE Training Code

[wojfran]

Hi,

Thank you for your excellent work on this repository. I noticed that the training code for VQVAE is not included, which makes it challenging to adapt and train the model on datasets beyond ImageNet.

I am aware of the vaex repository, which provides an implementation of a quantizer. However, as far as I understand, it doesn't include scaling, which differs from the VQVAE implementation here.

Would it be possible to provide the training code for VQVAE? This addition would greatly benefit researchers and practitioners who wish to implement the model on other datasets or explore its architecture in more detail.

Thank you for considering this request, and I appreciate the team's efforts in advancing the field!