---
license: apache-2.0
---
This dataset represents the source PDFs and rules used in  olmOCR-bench.

This benchmark is meant to easily evaluate OCR quality across a variety of models.