It would be nice if we could integrate Unity's ML-Agents toolkit as another way to train models for those who have forked the repository. Right now the only way to train models is through the Sorting Battle Gym.