PyTorch-GAN

PyTorch-GAN copied to clipboard

buffer in the cyclegan

1

My personal understanding about what the buffer do is that the first 50 fake sample from the generator are fed into the discriminator as usual, but in the meantime they...

SITUSITU

change the number of unlabeled data in semi-Supervised GAN

2

Hello It seems the code consider all data as labeled data. so how unlabeled data can be feed to the network? Thank you

sakh251

Typo in bicyclegan/models.py

2

https://github.com/eriklindernoren/PyTorch-GAN/blob/a163b82beff3d01688d8315a3fd39080400e7c01/implementations/bicyclegan/models.py#L153 in_channels -> channels (I guess)

Ice9Coffee

walter-234

Can I specify the number of images to be generated？

1

thank

aaawrong

Hi, i'm training image on 256x256. But, I how do I test on 512x512(as origianl image size)?

SYLin117

How to set training labeled sample number in SGAN

1

There doesn't have a ratio of unlabeled and labeled training data in the code.

Yerichen

some question about cyclegan.py

2

I'm a little confused about why use mse_loss in GAN loss and use L1 loss in Cycle loss and Identity loss,and i didn't find this in the paper. And the...

blight19

.sh 文件

1

.sh 文件里面的下载链接好像过期了，没法用了，不知道有哪位大佬那里有数据集啊（ The download link in the .sh file seems to be out of date and can’t be used anymore. I don’t know who has a data set there.）

starlitCoding

As I first touch the GAN, I wanna know how to use aae to data augmentation? Directly input an random matrix compling with Normal(0, 5) distribution into the decoder? or...

mia-7