All model names are simply the underlying class names at the moment.
It would be better to have a dictionary with short names: full name for each model, so that users can select models with short names.
There should also be a method get_models or similar that prints short name, long name, source
This should also resolve the problem that currently different PyTorch networks have the same name (NeuralNetTorch), which is why they can't be loaded at the same time
All model names are simply the underlying class names at the moment.
get_modelsor similar that prints short name, long name, source