dict形式のembeddingをgensimのbinで保存する．

save_gensim_embedding_from_dict_embedding.py

import gensim
import argparse

import numpy as np


def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument('--input', type=str, required=True)
    parser.add_argument('--output', type=str, required=True)
    args = parser.parse_args()

    return args


def save_embeddings(dict_embedding, output):
    emb_num = len(dict_embedding)
    emb_size = len(list(dict_embedding.values())[0])

    with gensim.utils.smart_open(f'{output}.bin', 'wb') as fw:
        fw.write(gensim.utils.to_utf8(f'{emb_num} {emb_size}\n'))
        for word, vector in (dict_embedding.items()):
            vector = vector.astype(np.float32).tostring()
            fw.write(gensim.utils.to_utf8(word) + b' ' + vector)


def main(args):
    dict_embedding = {}

    with open(args.input) as f:
        for l in f:
            word = l.split()[0]
            vector = np.array(l.split()[1:])
            dict_embedding[word] = vector
    save_embeddings(dict_embedding, args.output)


if __name__ == "__main__":
    args = parse_args()
    main(args)