search

lucidrains / stylegan2-pytorch

Simplest working implementation of Stylegan2, state of the art generative adversarial network, in Pytorch. Enabling everyone to experience disentanglement
https://thispersondoesnotexist.com
MIT License
3.67k stars 591 forks source link

is Dual Contrastive Loss useful？ #235

Open Johnson-yue opened 3 years ago

Johnson-yue commented 3 years ago

Hi， thank you provide a code for Dual Contrastive Loss， and is it useful for performance on FFHQ dataset？？

and did you add self-attention and reference-attention in D？

GLivshits commented 2 years ago

Hello. I don't know about the FFHQ, but on my domain, DCL worked better than Hinge loss.

Johnson-yue commented 2 years ago

@GLivshits which type of dataset ? human face ?

49xxy commented 2 years ago

Hi， thank you provide a code for Dual Contrastive Loss， and is it useful for performance on FFHQ dataset？？

and did you add self-attention and reference-attention in D？

May I ask if you have successfully used the loss？