PDF analysis. | OpenAI | Page 1

timid merlin Aug 15, 2023, 5:56 PM

#

I'm learning OpenAI api and I want to build project where I can upload PDF and use api to analyse it and get some outputs based on it. What are the steps to achieve it?
I'm familiar with react so that would be the easiest way to built UI for me.
I don't want to "chat with the pdf".
I want it for example to get some values and use them to fill the UI

ionic junco Aug 15, 2023, 6:12 PM

#

the API has no such fature, the part OAI services do are the aprt that handles the processing of the text, your system still has to somehow convert the PDF file into a text file

timid merlin Aug 15, 2023, 6:26 PM

#

So converting PDF to text outside of OAI, then using this text to make api call ? Event if the text will be large?

ionic junco Aug 15, 2023, 6:34 PM

#

the text can't be larger than the token limit of the model you use

timid merlin Aug 15, 2023, 7:20 PM

#

Is there a walk around if it is too large ?

grim night Aug 15, 2023, 8:56 PM

#

There's no easy way around trying to fit documents larger than the context window. You basically need to split the document and build a search system to find the right parts and assembed sources that fit within the context window.

#PDF analysis.