You can play with the code[0] for this really easy if you've got a Nvidia GPU with nvidia-docker[1] installed: docker run --gpus all -it anibali/pytorch:cuda

You can play with the code[0] for this really easy if you've got a Nvidia GPU with nvidia-docker[1] installed: docker run --gpus all -it anibali/pytorch:cuda-10.1 bash # copy images to container with `docker cp` pip install stylegan2_pytorch stylegan2_pytorch --data ./images [0] https://github.com/lucidrains/stylegan2-pytorch https://github.com/lucidrains/stylegan2-pytorch [1] https://github.com/NVIDIA/nvidia-docker https://github.com/NVIDIA/nvidia-docker