News for hackers. Stuff that matters.
Re:Language Models, Word2Vec, and Efficient Softmax Approxim... by rpedela on Tuesday October 31, 2017 @11:10PM (#15598562)
Each token would have one vector from word2vec. A token could be a word or phrase depending on the pre-processing. The words in a phrase are usually concatenated with an underscore. I recommend gensim if want/need phrases.
Re:Language Models, Word2Vec, and Efficient Softmax Approxim... by kirillkh on Wednesday November 01, 2017 @06:01AM (#15599887)
Ah, you're right, word2vec assigns one vector to each word, as opposed to one vector to each meaning. Then the problem remains: we can't differentiate between homonyms.

But it seems it's been solved, too: https://github.com/sbos/AdaGram.jl

Re:Language Models, Word2Vec, and Efficient Softmax Approxim... by rpedela on Wednesday November 01, 2017 @10:11AM (#15601253)
There is also sense2vec which I think tries to do something similar. https://explosion.ai/blog/sense2vec-with-spacy