Search code examples
pythondeep-learningnlpgensimword-embedding

pass from a model of type gensim.models.keyedvectors.Word2VecKeyedVectors to a model of type gensim.models.word2vec.Word2Vec


I downloaded a word embedding already train in "glove.txt" format I imported it in as a model of type gensim.models.keyedvectors.Word2VecKeyedVectors thanks to this documentation :

https://radimrehurek.com/gensim/scripts/glove2word2vec.html

But I would like a model of type gensim.models.word2vec.Word2Vec

Will there be a way to convert it or import it directly into the desired format?


Solution

  • A set of word-vectors isn't enough to create a full Word2Vec algorithm model, which includes a lot more information from training, including extra internal model weights & word-frequencies. (The word-vectors alone are less than half the state of the model.)

    Why do you want a full model rather than just the vectors?

    Can you train your own model from text data that's the same as, or similar is size/value to, the text used for creating the glove.txt word-vectors?