Implementation of the paper InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets by Xi Chen, Yan Duan, Rein Houthooft, John Schulman, Ilya Sutskever, Pieter Abbeel.
The implementation of the model is entirely based on the keras APIs. As of now, the model trainer uses tensorflow directly to create TensorBoard summaries (will be decoupled in the future).
Run with:
python main.py <experiment name>
Visualize in tensorboard:
tensorboard --logdir=<project root>