Is it possible to support PDF?

[C0ldSmi1e]

Hi, thanks for your work! I actually created a similar tool with Gemini. Recently I have to deal with some receipts in PDF format, and here's my solution: converting the PDF to images page by page, and send them to Gemini. I'd love to try llama in ocr.

[tushar-prox]

Since the code uses base64 for JPEG, it is possible to convert a PDF into a base64 JPEG and achieve the same result using pdf2pic

Something like #8

I am unable to test it locally to it's in draft state.

[pyprakhar]

Hi, I have done similar work with Gemini and I had used a approach for, getiing pdf's work.