JuliaText / Embeddings.jl

Functions and data dependencies for loading various word embeddings (Word2Vec, FastText, GLoVE)
MIT License
81 stars 19 forks source link

Paragram embeddings #9

Open oxinabox opened 6 years ago

oxinabox commented 6 years ago

I would like to add the Paragram embeddings http://www.cs.cmu.edu/~jwieting/

They were state of the art for somethings I care about, at one point fairly recently.

They are in the same plain text space delimited format as GLoVE. So the parse can just hand off to that one. So it is mostly writing the datadep strings

aquatiko commented 5 years ago

@oxinabox Is this still required to solve? If yes could you give some insights how to proceed?

oxinabox commented 5 years ago

It would be nice to have still yes. I have no insight on what is needed beyond look at the code that exist And added DataDep registrations to download it, and a parser if that is required

aquatiko commented 5 years ago

Should I add both Paragram-WS353 and Paragram-SL999 or the one in plain text format there?

oxinabox commented 5 years ago

Avoiding plaintext is generally preferable, for faster loading and smaller sizes.

tejasvaidhyadev commented 4 years ago

@oxinabox is this still required to solve ? I am thinking of adding both Paragram-WS353 and Paragram-SL999

oxinabox commented 4 years ago

Yes, it is still desirable