Sarath Shekkizhar

Hi @newvic2008 I trained for only 1e5 iterations which I believe is lot less compared to what the authors of the paper train the model for. Nevertheless, if it will...

The authors by discarding BN were trying to prove that WGAN formulation is much more stable - You are welcome to comment it out. I don't remember whether it was...

Hi @TheRevanchist It's been a while since I looked into the code, but if it is implemented with padding I think you are right in that this was a mistake...

@Sora77 I assume you are talking about the context encoder with face implementation. Yes I had used only reconstruction loss for the problem. Sorry for not making it clear what...

Hi @hiteshvaidya, I wrote these code for experimenting with different ideas when tensorflow was fairly young. I'm aware of changes in tensorflow but haven't really got around to updating all...

@AntonioCarta - The issue I was describing is that the resize operation for a single addition of new samples is done thrice - as pointed out in the code snippet....

Possible related issue https://github.com/shekkizh/WassersteinGAN.tensorflow/issues/3 Let me know if the solution posted in the linked issue solves the problem.