AC-GAN

A basic implementation of Auxiliary Classifier Generative Adversarial Network (ACGAN) in Torch7 on the MNIST dataset by Odena et al.

##Basic Requirements: Torch

CuDNN (not required but recommended)

Run as:

th main.lua

To train on gpu:

th main.lua --gpu=1

Sample images will be stored in samples directory. Both the generator and the discriminator models are saved in the checkpoints folder after each epoch.