0
$\begingroup$

I have the pre-trained Embbedings on the language. I have the vocabulary for that language, what would be the pipeline to train this vocabulary by using Pre train embeddings through the word2vec model?

$\endgroup$

1 Answer 1

0
$\begingroup$

Are you doing a word2vec fine-tuning?

If yes, here is a guide on how to do it:

https://czarrar.github.io/Gensim-Word2Vec/

See also: https://www.kaggle.com/code/rtatman/fine-tuning-word2vec/notebook

$\endgroup$
5
  • $\begingroup$ I have pre-train word embeddings from this link link and the vocabulary of the language with unique words, do I need to train it again on the vocabulary? or is it possible to train it again? $\endgroup$ Commented Jan 20, 2023 at 9:29
  • $\begingroup$ Do you mean with a new vocabulary in addition to the existing one? $\endgroup$ Commented Jan 20, 2023 at 9:59
  • $\begingroup$ I would say the vocabulary is going to same because that contains the dictionary of the particular language. Do we still need to train it again, or I can compute the embeddings using the pre-train embeddings? Thank you for the reply :) $\endgroup$ Commented Jan 20, 2023 at 10:09
  • $\begingroup$ Thank you I only need to compute the embeddings through the pre-trained embeddings The fine tune i required when the data is new or of other language, I am new to NLP that's why $\endgroup$ Commented Jan 20, 2023 at 13:34
  • 1
    $\begingroup$ That's why fine-tuning seems more appropriate: you reuse the model already trained, and you train it again with new data. The links above should do the job. $\endgroup$ Commented Jan 20, 2023 at 13:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.