OCR Tools can't be used out of the box

[samuelswandi]

I've deployed my version of olmOCR through the https://deepinfra.com/allenai/olmOCR-7B-0725-FP8's private deployment. While using the python script, there's a bunch of missing requirements that stopped me to use it right away:

1. pypdf

2. olmOCR

I think the required should be

olmocr==0.2.1
openai==1.98.0
pypdf==5.9.0

Additional:

• I only know that this exist because of tweet on X (https://x.com/DeepInfra/status/1951836336585113869), there's no README/notes to know how to attach pdf/image as the request