3 ms·
It doesn't seem to cover contextual word embeddings though which are all the rage since BERT came in.
by omegalulw 5y ago
It doesn't seem to cover contextual word embeddings though which are all the rage since BERT came in.
- gillesjacobs 5y agoYou're right, I too was a bit disappointed that the page didn't mention contextual embeddings since that has been the best performing and most robust approach in NLP for nearly four years now. It's still a clear introduction to word vectors though and it is good to know where we came from.