Obtain embedding vectors

[rominaappierdo]

Hello and thank you for sharing your work!
I would like to know if there is a way to obtain embedding vectors of one (or more) sentences fed into the model.
Hope you could help me.
Thank you in any case

[MichalBrzozowski91]

Hi, thanks for feedback. Can you elaborate what is your use case? So far we have implemented support for longer sequences for the binary classification problem.

[rominaappierdo]

Thank you for your reply, I just wanted to know if there was a way to simply obtain embedding vectors. I would like to just obtain a representation vector for a given (longer) sequence, no need to perform downstream tasks currently. Is it possible to obtain them with your work?

[MichalBrzozowski91]

Hi, I was working on it and I think it is possible to obtain embedding vectors using our code. I created a PR with example notebook.
Feel free to ask any questions.
I hope it will help you.

[MichalBrzozowski91]

HI, we made major changes in this repo. Now in order to obtain embedding vectors, we suggest getting the tokenized chunks from the method _tokenize of the class BertClassifierWithPooling. The obtained tokens can then be directly fed to the pre-trained BERT to obtain embeddings of each chunk.