Reputation: 1113
I have followed this documentation to get fasttext running, fairly straightforward. Commands such as $ ./fasttext nn result/fil9.bin
can be used to find nearest neighbour of a word. However, is there any such simple command in fasttext that takes two words such as "Vehicle" and "Car" as input and returns their (cosine) similarity i.e. a number like 0.777.
Upvotes: 1
Views: 787
Reputation: 13746
At present time, it's not part of the API. You can use gensim, or write your own distance metric calculation.
Upvotes: 1