samples.md

Training on the MNIST dataset

GANs

• DCGAN (sigmoid activation on the last layer)
• MAD-GAN
• WGAN and GoGAN
• BEGAN

Datasets

• MNIST

Comments

• All of the GANs are trained with a fixed LR=1-e5 for 100k iterations.
• Disclaimer: no hyper-parameter search has been done yet.

Generated samples

DCGAN / After 5k, 50k, 99k iterations

MADGAN / After 5k, 50k, 99k iterations

WGAN / After 5k, 50k, 99k iterations

GoGAN / Stage 1 after 5k, 50k iters and stage 2 after 98k iters

BEGAN / After 5k, 50k, 99k iterations