11 lines
520 B
Markdown
11 lines
520 B
Markdown
|
# StyleGAN Implementations
|
||
|
|
DLAS supports two different StyleGAN2 implementations:
|
||
|
|
|
||
|
|
- [@rosinality implementation](https://github.com/rosinality/stylegan2-pytorch/commits/master)
|
||
|
|
Designed to reach parity with the nVidia reference implementation in TF1.5
|
||
|
|
- [@lucidrains implementation](https://github.com/lucidrains/stylegan2-pytorch)
|
||
|
|
Designed with simplicity and readability in mind.
|
||
|
|
|
||
|
|
I prefer the readability of @lucidrains implementation, but you cannot (yet) use pretrained weights
|
||
|
|
with it. I'm working on that.
|