fastencoder

fastencoder

Use pretrained fasttext models to generate word and sentence embeddings.

Data

Download word embeddings here.

Create data.vec to be used by application:

head -n 50000 `filename`.vec > data.vec

Usage

let FastEncoder = require('fastencoder');

(async () => {

    // Instantiate and load encoder
    let encoder = new FastEncoder();
    await encoder.load('./data.vec');

    // Get word/sentence embedding
    let vector = encoder.encode('hello');
    console.log(vector.length);

    // Get most similar words
    let similar = encoder.getSimilar('hello', 3);
    console.log(similar);

})();