DocParsingBench — Open Data Commons Attribution License v1.0 (ODC-BY 1.0)

Source: https://huggingface.co/datasets/SoMarkAI/DocParsingBench
Copyright: SoMark.
Suites: docparsing_images, and the synth_copies pages rebuilt from it.

Page images are used as shipped upstream; their annotations are not used, and this benchmark's tests
were made for it. Upstream describes the documents as drawn from real business workflows.

Please cite: SoMark (2026). DocParsingBench. Hugging Face and ModelScope.

Full license text: https://opendatacommons.org/licenses/by/1-0/
