Ask for the training curves

[sangwoomo]

Hi, I tried to reproduce your results on MNIST dataset, but the model seems to diverge.
I simply clone your repo and run python train.py --dataset Mnist.

Here are my loss curves and generated samples.

Can you share your loss curves and generated samples during training?
(I'm using PyTorch v1.1)

[sangwoomo]

It seems to work well after thousands of epochs.

[LuEE-C]

The training curves just mean that the discriminator is getting better and better in relation to the generator's performance and do not necessarily imply divergence in a gan. I do not have access to training curves right now but this seems right to me. Do note that it takes 10's of thousands of steps to produce good results