4 ms·
Thank you. Yes, exactly, that's my sense too (pun intended) :-) Naturally, I'm wondering whether it might be possible somehow to approximate a (pretrained) EL
by cs702 8y ago
Thank you. Yes, exactly, that's my sense too (pun intended) :-)
Naturally, I'm wondering whether it might be possible somehow to approximate a (pretrained) ELMo (or similar) model with two simpler transformations: first a transformation to the space of word-sense compositions (e.g., via GloVe/SGNS), and then a transformation to a space that somehow encodes probabilities over word senses given the context. Hmmm...
- thesehands 8y agoThere may be merit in tagging each of the words with their part of speech prior to fitting the model in a similar way to sense2vec. Using your example above you would then have 2 vectors, one for leaves|VERB and one for leaves|NOUN