Jon Harmon
Jon Harmon
Did anything come of this? I'm trying to follow the breadcrumbs in various repo discussions 🙃
This should happen `.onLoad` of the package.
Very much unpaused, for anybody following along.
Note: I tested preprocessing the config json vs saving it as-is, preprocessing saves microseconds, so it probably isn't worth messing with. It wouldn't HURT, though, so I may do the...
This could probably relate to #30.
I *think* the best option here would be to make a function, something like `known_models`. *It* would have defaults and information about what models we know how to deal with....
bert_tiny_uncased_128 is a 15.7MB zip and ~works. That's small enough that github will allow it without lfs.
Hmm, but I think maybe we download at the start now anyway? DLing this one would be way faster!
Update: Almost definitely! We're finally actively working on this again, and the path will definitely be along the lines of "get the TF2 version working, then investigate {torch}."
It's likely that we'll make a separate repo for torch development (although that isn't 100% decided yet). I'll definitely ping you when we figure it out!