Implementing the Boundary-Seeking GAN in TensorFlow. Credit to the authors of the original paper, who also released their Theano implementation.
First, download the celebrity faces (CelebA) dataset, put it in data/
, and unzip it such that we have data/img_align_celeba/
.
Then prepare the tfrecords:
python make_tfrecords.py
Set training params in train.py and run training with:
python train.py