平均多个模型Word2vec Gensim

Question

I have trained few million words in Word2Vec of Gensim of Python. I want to update this trained model with new data. But from your previous posts and other sources around the web I came to know this is not possible. So I am trying to create multiple models and dump them. Now I want to merge the models I am dumping. I want to use these dumped results. I got a previous post Merging pretrained models in Word2Vec? but I am not getting how to do it. I came to know there is a library named deepdist, I am trying to see some experiments around:

model = word2vec.Word2Vec.load_word2vec_format('/tmp/vectors.bin', binary=True)

1. Is there a possible solution?
2. If any, one may kindly suggest how to do it?

I am using Python2.7 on Windows 7 Professional.

Answer 1

The answer you pointed to does NOT suggest merging the models as a solution. Actually, they are suggesting you to use the different models that you have separately. Cast a prediction with each of the models and then combine the answers. There are several approaches to combine the output. In your case you mention you have several models, so you can ignore the part of that answer where they suggest to break your training data in 2 in order to actually have 3 models casting predictions. You can use a majority voting policy as long as you have more than 2 predictions.