MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scale
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918436296196096 |
|---|---|
| author | Wang, Bin He, Tianyao Ouyang, Linke Wu, Fan Zhao, Zhiyuan Chu, Tao Qu, Yuan Jin, Zhenjiang Zeng, Weijun Miao, Ziyang Xu, Bangrui Niu, Junbo Cai, Mengzhang Qiu, Jiantao Zhang, Qintong Ma, Dongsheng Sun, Yuefeng Dong, Hejun Zhang, Wenzheng Xiao, Jutao Shi, Jiayong Liao, Pengyu Zhao, Xiaomeng Zhong, Huaping Wei, Liqun Yu, Jing Yang, Jie Li, Wei Wang, Shasha Wu, Qianqian Zhou, Xuanhe Li, Weijia Li, Zhenxiang Tu, Zhongying Wu, Jiang Wu, Lijun Xu, Chao Chen, Kai Zhang, Wentao Qiao, Yu Zhou, Bowen Lin, Dahua He, Conghui |
| author_facet | Wang, Bin He, Tianyao Ouyang, Linke Wu, Fan Zhao, Zhiyuan Chu, Tao Qu, Yuan Jin, Zhenjiang Zeng, Weijun Miao, Ziyang Xu, Bangrui Niu, Junbo Cai, Mengzhang Qiu, Jiantao Zhang, Qintong Ma, Dongsheng Sun, Yuefeng Dong, Hejun Zhang, Wenzheng Xiao, Jutao Shi, Jiayong Liao, Pengyu Zhao, Xiaomeng Zhong, Huaping Wei, Liqun Yu, Jing Yang, Jie Li, Wei Wang, Shasha Wu, Qianqian Zhou, Xuanhe Li, Weijia Li, Zhenxiang Tu, Zhongying Wu, Jiang Wu, Lijun Xu, Chao Chen, Kai Zhang, Wentao Qiao, Yu Zhou, Bowen Lin, Dahua He, Conghui |
| contents | Current document parsing methods advance primarily through model architecture innovation, while systematic engineering of training data remains underexplored. Yet state-of-the-art models spanning diverse architectures and parameter scales exhibit highly consistent failure patterns on the same set of hard samples, suggesting that the performance bottleneck stems from shared deficiencies in training data rather than from architectural differences. Building on this finding, we present MinerU2.5-Pro, which advances the state of the art purely through data engineering and training strategy design while retaining the 1.2B-parameter architecture of MinerU2.5 unchanged. At its core is a Data Engine co-designed around coverage, informativeness, and annotation accuracy: Diversity-and-Difficulty-Aware Sampling expands training data from under 10M to 65.5M samples while mitigating distribution shift; Cross-Model Consistency Verification leverages output consensus among heterogeneous models to assess sample difficulty and generate reliable annotations; the Judge-and-Refine pipeline improves annotation quality for hard samples through render-then-verify iterative correction. A three-stage progressive training strategy--large-scale pre-training, hard sample fine-tuning, and GRPO alignment--sequentially exploits these data at different quality tiers. On the evaluation front, we rectify element-matching biases in OmniDocBench v1.5 and introduce a Hard subset, establishing the more discriminative OmniDocBench v1.6 protocol. Without any architectural modification, MinerU2.5-Pro achieves 95.69 on OmniDocBench v1.6, improving over the same-architecture baseline by 2.71 points and surpassing all existing methods, including those based on models with over 200x more parameters. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_04771 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scale Wang, Bin He, Tianyao Ouyang, Linke Wu, Fan Zhao, Zhiyuan Chu, Tao Qu, Yuan Jin, Zhenjiang Zeng, Weijun Miao, Ziyang Xu, Bangrui Niu, Junbo Cai, Mengzhang Qiu, Jiantao Zhang, Qintong Ma, Dongsheng Sun, Yuefeng Dong, Hejun Zhang, Wenzheng Xiao, Jutao Shi, Jiayong Liao, Pengyu Zhao, Xiaomeng Zhong, Huaping Wei, Liqun Yu, Jing Yang, Jie Li, Wei Wang, Shasha Wu, Qianqian Zhou, Xuanhe Li, Weijia Li, Zhenxiang Tu, Zhongying Wu, Jiang Wu, Lijun Xu, Chao Chen, Kai Zhang, Wentao Qiao, Yu Zhou, Bowen Lin, Dahua He, Conghui Computer Vision and Pattern Recognition Computation and Language Current document parsing methods advance primarily through model architecture innovation, while systematic engineering of training data remains underexplored. Yet state-of-the-art models spanning diverse architectures and parameter scales exhibit highly consistent failure patterns on the same set of hard samples, suggesting that the performance bottleneck stems from shared deficiencies in training data rather than from architectural differences. Building on this finding, we present MinerU2.5-Pro, which advances the state of the art purely through data engineering and training strategy design while retaining the 1.2B-parameter architecture of MinerU2.5 unchanged. At its core is a Data Engine co-designed around coverage, informativeness, and annotation accuracy: Diversity-and-Difficulty-Aware Sampling expands training data from under 10M to 65.5M samples while mitigating distribution shift; Cross-Model Consistency Verification leverages output consensus among heterogeneous models to assess sample difficulty and generate reliable annotations; the Judge-and-Refine pipeline improves annotation quality for hard samples through render-then-verify iterative correction. A three-stage progressive training strategy--large-scale pre-training, hard sample fine-tuning, and GRPO alignment--sequentially exploits these data at different quality tiers. On the evaluation front, we rectify element-matching biases in OmniDocBench v1.5 and introduce a Hard subset, establishing the more discriminative OmniDocBench v1.6 protocol. Without any architectural modification, MinerU2.5-Pro achieves 95.69 on OmniDocBench v1.6, improving over the same-architecture baseline by 2.71 points and surpassing all existing methods, including those based on models with over 200x more parameters. |
| title | MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scale |
| topic | Computer Vision and Pattern Recognition Computation and Language |
| url | https://arxiv.org/abs/2604.04771 |