Jun-Yan Zhu

It shows that you DATA_ROOT : "" . Have you set the DATA_ROOT?

Setting hyper-parameters is often a black magic in deep learning models. There is probably no principled way of doing it. In practice, we find that `lambda=10~100` yields good results.

I fixed the bugs in the `save_current_results`. Let me know if it works for you with batchSize>1

@DmitryUlyanov Thanks for the notes. @Quasimondo I guess the tiling operation will still cause the boundary artifacts even without instanceNorm. Let's say your images are 512x512. I would recommend that...

Thanks for your fixes. There might be some issues about visibility test in the code. I haven't had a chance to fix them.

You need to install it via `sudo pip install qdarkstyle`.

Very interesting. Are you using conda? For conda users, you can try [this](https://anaconda.org/auto/qdarkstyle).

Please see the [requirements](https://github.com/junyanz/iGAN#requirements)

It seems to be related to Theano. Could you run other Theano [example](https://github.com/Newmu/Theano-Tutorials)/test code? You may want to install this Theano [version](https://github.com/Newmu/dcgan_code/issues/5). The code was developed around 2016, and the...

In `THEANO_FLAGS`, you can use `device=cpu` rather than `device=gpu0`.