-
Hello!
When I use formula for loss of WGAN I get negative value for such loss, how did you you avoid this?
Thanks!
-
I am trying to implement WGAN-GP with cntk, but here comes a big difficulty.
Suppose x is an input variable and F is the discriminator neural network, then first we need to get the gradient over x…
-
https://github.com/wiseodd/generative-models/blob/c790d2c8add3d600c03ee9a301a969e10ccd7562/GAN/wasserstein_gan/wgan_tensorflow.py#L82-L83
in the code , you define the discriminator loss is :
`D_loss…
-
I would like to know how to train WGAN-GP with variable batch sizes. I couldn't find a way to set the batch size in the function below dynamically. This is needed because my last batch is not complete…
-
Everything is so much nicer in TensorFlow. No CPU/GPU memory swapping for FFTs, for example.
Stay tuned.
-
Hi, excellent work! I have a question that in origin WGAN paper and its code, in one iteration discriminator is trained several sub-iterations before generator is trained, but that's not the case in y…
chmxu updated
6 years ago
-
Should the following https://github.com/eriklindernoren/PyTorch-GAN/blob/a163b82beff3d01688d8315a3fd39080400e7c01/implementations/wgan_gp/wgan_gp.py#L162 be this instead:
```
fake_imgs = generator(…
-
Hi,
I was wondering if anyone has tried to use wgan_gp for the gan mode?
I'm not sure how to implement the gradient penalty here, any help is appreciated.
-
Hi, I was wondering why did you guys decide to use batch norm in a critic of WGAN-GP. The paper on Improved training of WGAN(the one where gradient penalty is proposed) advises against it.
Thanks in…
-
File "C:/Users/Daxika/Desktop/improved_wgan_training-master/improved_wgan_training-master/gan_toy.py", line 239, in
_data = gen.next()
AttributeError: 'generator' object has no attribute 'nex…