Spaces:
Running
Please add support for multiple PDFs
please. at least any guidance? I am pretty new to this
Hi
@Pixel69
,
A high level architecture is given here:
https://github.com/bhaskatripathi/pdfGPT
I have added your request for multiple PDF support and will develop it over the weekends. Will soon share an update of the new release.
nice, thanks!
I was also playing with the script and here's what I found:
you could calculate the hash of the uploaded pdf, create a dict with hash as the key, and for the value pass it the SemanticSearch instance, next time if the same pdf is uploaded the script does not need to "load_recommender" again.
And I saw it was using only my CPU, any way for it to use my GPU?
yeah load_recommender can be improved. I am planning to integrate pinecone.
GPU will not be able to boost performance here (unless one runs it on local) as we are hitting openAI API and calling Universal sentence encoder from HF.