Closed schwartzadev closed 5 years ago
Hey - Thank you! Most of the effort for this project is continuing in the cx_db8_flair.py file instead, as I found that to be a much easier framework to work with for having access to lots of embeddings.
That being said, Flair doesn't support the Universal Sentence Encoder right now, so the file you refactored is still very useful (I've seen some stuff indicating that the Universal Sentence Encoder is the best contextual pretrained word embedding set right now) - so I will likely merge this and try to get the same code improvements in cx_db8_flair
Hopefully this makes this file easier to understand!
Let me know if you have any questions 👍