Skip to content

Conversation

@justinjohn0306
Copy link

No description provided.

@johndpope
Copy link

johndpope commented Jan 7, 2021

the customops was bumped by nvidia - for 2020 here / along with a bunch of other dnnlib functions.
https://github.com/NVlabs/stylegan2-ada/blob/main/dnnlib/tflib/custom_ops.py

N.b. - nvidia are abandoning tensorflow and going to pytorch - I have a branch that supports tensorflow2
https://github.com/johndpope/stylegan2-ada/

more elaborate / cherry picked branch here with training support.
https://github.com/johndpope/stylegan2-ada/tree/digressions

@KINOBOT
Copy link

KINOBOT commented Jan 8, 2021

Hey, I am trying to replicate this, and have used your PR, started from scratch, and edited the custom_ops.py file.

After that, I still get
AssertionError: Torch not compiled with CUDA enabled
which must be because _pytorch_select=1.1.0=cpu is in your env? Since then, I have rerun
conda install --force-reinstall pytorch==1.1.0 torchvision==0.3.0 cudatoolkit=10.0 -c pytorch to reinstall the torch version as per the paper. Now it runs. We just need to fix the UI for windows now ;)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants