Save and load custom trained model with HF model hub

[BramVanroy]

I am incredibly excited by NLP tools working together and integrating. HuggingFace's model hub is a nice central environment to keep track of models of all kinds - not even limited to models of their own libraries.

With the 1.3 release, stanza now also makes use of this model hub. That's awesome! For now, I do not see the integrated option to push your own custom trained models to this hub from within stanza, nor can you easily load custom models from the hub (I think?). I think it would be a great addition. That would for instance allow users to load a model from the hub like so.

model = stanza.Pipeline("funky-user/flashy-custom-model")

Inspiration may be found in spaCy's implementation.

I kind of suspect this feature is already being worked on, but I'll post it here as a show of support and excitement!