Simple implementation of DCGAN [Paper], see also: [cs231n]
python train_dcgan.py
MNIST Generation results after 15 epochs:
Pytorch implementation of the Wasserstein gan. [Paper]
python train_wgan.py
MNIST Generation results after 3000 iterations:
Pytorch implementation of VAEGAN. [Paper]
python train_vaegan.py --dataset-name CelebA --data-dir /path/to/noncropped/CelebA/
CelebA generation after 1 epoch:
Cropped CelebA generation after 4 epochs:
Adversarial Auto Encoder [Paper]
python train_aae.py --model-type MNIST --dataset MNIST --epochs 10 --latent-size 2
Latent manifold visualisation after 10 epochs:
