/vidgen/generative-models/configs/example_training/autoencoder