GloVe txt to gensim word2vec model.py

import numpy
from gensim.models.keyedvectors import Vocab, KeyedVectors

# Covert the Twitter datasets found here to gensim word2vec format: https://github.com/3Top/word2vec-api

with open('glove.6B.300d.txt') as fin:
    word_vecs = {
        word: numpy.array(list(map(float, vec.split())))
        for line in fin
        for word, vec in [line.split(' ', 1)]
    }
    
    result = KeyedVectors()
    result.vector_size = matrix.shape[1]
    result.index2word = word_vecs.keys()
    result.syn0 = numpy.concatenate([word_vecs[k][None, :] for k in result.index2word], axis=0)
    del word_vecs

    for word_id, word in enumerate(result.index2word):
        result.vocab[word] = Vocab(index=word_id, count=matrix.shape[0] - word_id)

result.save_word2vec_format('glove.6B.300d.bin', binary=False)