This repository contains the implementation of COEGAN and all code used in the evaluation and comparison with other methods, as presented in the paper COEGAN: Evaluating the Coevolution Effect in Generative Adversarial Networks http://gecco-2019.sigevo.org.

Install pytorch:

conda install pytorch torchvision cuda90 -c pytorch

Install dependencies:

pip install -r requirements.txt

python -m unittest discover

Edit the experimental settings in evolution/config.py.

python ./train.py

Run JupyterLab

jupyter lab

See below the results of the experiments presented in the paper:

https://stackoverflow.com/questions/55124407/output-and-broadcast-shape-mismatch-in-mnist-torchvision change this line in /evolution/gan_train.py:

transform = transforms.Compose([transforms.ToTensor(),
  transforms.Normalize((0.5,), (0.5,))
])