Closed PJ-Finlay closed 3 years ago
It looks like they reduced the distribution size which should be nice for the Mac app which uses seq2seq sentence boundary detection. Stanza is ~700MB with some crazy Nvidia binary that comes with PyTorch so it probably won't matter too much for the main Python package.
Done, very straightforward!
Notes:
We use translation scores and will need to start setting return_scores to true.
Improve GPU docs
Python docs