How to using PDF file as input type?

[JonWangM]

Will the team have a PDF as input to the development plan?

[akashAD98]

as of now its supporting text/csv . you can convert your pdf into text file & it ill work

[JonWangM]

As we know, PDF contains a variety of data information, such as tables, pictures, citations, chapters, etc., if we convert to txt, will we lose a lot of information?

[l4b4r4b4b4]

I use unstructured to ingest pdfs and construct txt files in input.

[AlexCastroAlex]

Hello , Did you use API deployed or front end ?
And where did you edit code to use unstructured ?
That's a real lack that a graph rag can not ingest many types of files like AI Search can 😢

[l4b4r4b4b4]

i used unstructured library if thats what you mean with the first question.

I did not change any graphrag code. Simply used unstructured to create txt from pdfs.

[AlexCastroAlex]

Well ok i see.
Thanks for answer 😊

[yaroslavyaroslav]

Batch images (of pdf) processing with gpt-4o-mini costs me around 1$ for everything (a single yet complex tech book of 300 pages) and provide quite impressive result.