Using set encoders to encode Wasserstein Spaces. Code and experiments for the paper "Permutation Invariant Networks to learn Wasserstein metrics" (Spotlight at Neurips TDA workshop https://tda-in-ml.github.io/cfp) https://arxiv.org/abs/2010.05820 #TODO 1) Try out wider networks. 2) Try other PINNs 3) Clean up code.