Can I use TransformerDocumentEmbeddings to generate embeddings of publications abstracts ?

[pruthvi-hegde]

I am working on a usecase where I need to generate document embeddings of abstracts of few articles. I am using TransformerDocumentEmbeddings to instantiate PubMedBert model. And, it generates 768 dimensional vector per document. Since, flair documentation says that the input to TransformerDocumentEmbeddings is a sentence, is it fair to feed in paragraph of multiple sentences and obtain embeddings this way ?

[whoisjones]

Yes you can obtain your embeddings this way, flair's Sentence object does not necessarily end with "."

[pruthvi-hegde]

Thank you @whoisjones

Is there any documentation that I can refer to understand the internal working of 'TransformerDocumentEmbeddings' class ? Basically I want to know how 'TransformerDocumentEmbeddings' calculates embedding for a given paragraph.

Thanks in advance!

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.