StyleGan2 model trained with official
StyleGan3.
But I modified the code (networks_stylegan2.py and dataset.py) to support non-square resolutions.
Note: Fp16 version (fbanime.pkl) only works on gpu. And fp32 version works on gpu and cpu.
e4e model trained with
encoder4editing-stylegan3.
I add support for official StyleGan2 model and change backbone to ResNet-34 in
restyle-encoder.
YOLOv5 model trained with official
YOLOv5