serge
serge copied to clipboard
Add the possibility to upload PDFs in the front part of the application to read it
Create on the front end the possibility to upload a PDF file for example with the following procedure:
- Parsing the file to read it
- Split content to respect "limit" of tokens for model to send.
- To be able to make queries on the elements of the document from the prompt.
(Of course it's not critical but it would really be a plus on the front)
Yeah that would be super useful agreed! I think we need to implement a LangChain integration (see #4) in order to do this the easy way and not reinvent the wheel.
I will try to work on this in the near future
I'll look into the LangChain integration on my end, if I can help.
For sure, feel free to share your findings
The models usually can't actually read very much without forgetting what they already read. You might be able to upload a page or two, but for that much you could probably copy and paste just as easily.