stylegan2-pytorch
stylegan2-pytorch copied to clipboard
rosinality
conditional implementation
Hi, did you also implement a conditional implementation, so that I can pass files with labels, as in the original TensorFlow implementation?
Currently this implementation does not support label supervision. But it will be not very hard to add it.
Hi @rosinality , thank you for your answer. Okay then, I will try to add it (and let you know if I manage). If in the meanwhile you add it, please keep me updated! :)
Hey, I forked your repository and made a fast (and not very general) implementation yesterday. I am still checking if it converges on my dataset, but maybe you may like to give a look at it. :)
https://github.com/PieraRiccio/stylegan2-pytorch
Nice work! Definitely it looks like how this implementation should have been.
@PieraRiccio Sorry for resurrecting this. I'm trying to convert a conditional model trained using Nvidia's implementation (from https://github.com/NVlabs/stylegan2-ada-pytorch ) to rosinality's format.
Am I correct in assuming you followed the same implementation of label conditioning as the NVIDIA implementation? Thank you!
Hey there! The PyTorch implementation for the conditional training did not exist at that time, but I replicated the implementation in TensorFlow, so I guess the answer to this question is: yes! :)