# Baidu releases MarathonDocBench for 100-300-page document parsing

> Original title: Baidu's MarathonDocBench evaluates the parsing of 100–300-page Chinese and English documents, including cross-page text, tables, and docu...

- Source: ModelScope
- Published: 2026-11-04T12:40:22.000Z
- Bad Influence: https://badinfluence.news/items/e83gfjw540coh5az8lok9x6ak
- Original: https://x.com/ModelScope2022/status/2117960142511813062

## Summary

Baidu 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.

## Post

Baidu's MarathonDocBench evaluates the parsing of 100–300-page Chinese and English documents, including cross-page text, tables, and document structure. 🤖https://modelscope.ai/datasets/baidu/MarathonDocBench

📚 Hundred-page scale: 100 documents, approximately 18,000 pages, and 205,000 annotated layout elements.

🔗 Structure across pages: Human-refined annotations cover bounding boxes, heading hierarchy, reading order, and cross-page text and table continuations.

🧪 Document-level evaluation: Reuses MPDocBench-Parse's eight metrics to assess text, tables, formulas, reading order, figures, and heading structure. Overall is their mean.

📊 A demanding test: Listed general-purpose VLMs score below 34 Overall under the published whole-document protocol, which counts failed requests as empty predictions. Input and output limits also affect results.

📜 Apache-2.0 License.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWSBaxObYAAdji8.png&mode=thumb&exp=1794009600&sig=59c530e07d94becd)

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWSBbl-awAAoSG5.jpg&mode=thumb&exp=1794009600&sig=0a414dfc68aac4c4)