This is an implementation of a Deep Convolutional Generative Adversarial Network (DCGAN) in TensorFlow. The example is based on the original work from Radford et. al. , where the generator learns to reconstruct samples from the MNIST dataset.
Tested on:
- macOS - CPU - Python 3.5 - Tensorflow 1.1.0
- ubuntu xenial - GPU - Python 3.5 - Tensorflow 1.1.0
Digit generations are as following after 5000 and 20000 iterations:
