Baidu releases MarathonDocBench, a benchmark for 100–300-page document parsing
Overview
Baidu releases MarathonDocBench, a benchmark for parsing Chinese and English documents of 100 to 300 pages, under an Apache-2.0 License, according to ModelScope's post on X. The benchmark contains about 18,000 pages and 205,000 annotated layout elements, with human-refined annotations for bounding boxes, heading hierarchy, reading order, and text and table continuations across pages.
Under the published whole-document protocol, which counts failed requests as empty predictions, listed general-purpose vision-language models score below 34 Overall.
Written by AI from the articles below · updated Nov 4, 7:48 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
ModelScope@ModelScope2022OfficialBaidu releases MarathonDocBench for 100-300-page document parsingAIBaidu releases MarathonDocBench, a benchmark for parsing 100–300-page Chinese and English documents, with about 18,000 pages and 205,000 annotated layout elements. Human-refined annotations cover bounding boxes, heading hierarchy, reading order, and cross-page text and table continuations. Listed general-purpose VLMs score below 34 Overall under the published whole-document protocol, which counts failed requests as empty predictions.

Heat Trend
Not enough continuous observations to show a trend yet.