Word2Vec：使用 Gensim 上传预训练的 word2vec 文件时收到错误

Question

I receive an error when trying to upload a pre-trained word2vec file (compiled with fasttext) using Gensim. File has '.vec' extension and can be found here: http://89.38.230.23/word_embeddings/we/corola.300.20.vec.zip

What I've tried so far: Option 1: KeyedVectors from gensim.models Option 2: FastText wrapper

#Option 1
    from gensim.models import KeyedVectors
    model = KeyedVectors.load_word2vec_format('Word_embeddings/corola.300.20.vec', binary=True)
######

#Option 2
    from gensim.models.wrappers import FastText
    model = FastText.load_word2vec_format('Word_embeddings/corola.300.20.vec')

Error option 1: UnicodeDecodeError: 'utf-8' codec can't decode byte 0x9b in position 0: invalid start byte

Deprecation Error option 2: DeprecationWarning: Deprecated. Use gensim.models.KeyedVectors.load_word2vec_format instead.

I need the correct method to successfully upload the word2vec file, using gensim.

Thank you.

Answer 1

Sometimes, it's just fine to use the unicode_errors='ignore' parameter, since there can be errors in the word embedding file. Just try:

model = KeyedVectors.load_word2vec_format('Word_embeddings/corola.300.20.vec', binary=True, unicode_errors='ignore')