Infinity-Parser2 Technical Report
Quick Answer
Infinity-Parser2 is a multimodal model that integrates a data-synthesis pipeline with multi-task reinforcement learning for document parsing.
Quick Take
It features a 5M bilingual corpus and achieves state-of-the-art results on benchmarks like olmOCR-Bench (87.6%) and ParseBench (74.3%), outperforming competitors like DeepSeek-OCR-2 and PaddleOCR-VL-1.5.
Key Points
- Developed a scalable synthesis engine for creating annotated document corpora.
- Infinity-Doc2-5M includes 5 million bilingual samples with diverse document types.
- Introduced a multi-task reward system for joint reinforcement learning across eight objectives.
- Infinity-Parser2-Flash offers 3.68x throughput improvement for low-latency inference.
- Infinity-Parser2-Pro achieves state-of-the-art performance on multiple parsing benchmarks.
Paper Resources
📖 Reader Mode
~2 min readAuthors:Zuming Huang, Jun Huang, Kexuan Ren, Baode Wang, Weizhen Li, Jianming Feng, Yu Wang, Yichen Yao, Shijun Lin, Yige Tang, Cheng Peng, Weidi Xu, Wei Chu, Yinghui Xu, Yuan Qi
Abstract:We present Infinity-Parser2, a large multimodal model that couples a controllable data-synthesis pipeline with multi-task reinforcement learning for end-to-end document parsing, addressing the persistent scarcity of faithfully annotated parsing corpora. Our contributions are threefold. First, we build a scalable synthesis engine, pairing a controllable rendering framework with an iterative refinement loop, and use it to construct and open-source Infinity-Doc2-5M: a 5-million-sample bilingual (Chinese/English) corpus spanning diverse document types, annotated with element bounding boxes, canonical content forms (Markdown, HTML, LaTeX, SMILES, structured charts), and full-page reading order. Second, we introduce a verifiable, multi-task reward system that enables Joint Reinforcement Learning across eight co-trained objectives (document parsing, layout analysis, table parsing, math formula parsing, chart parsing, chemical formula parsing, document VQA, and general multimodal understanding), unifying perception, structure, and reasoning in a single optimization signal. Third, we release two variants under a shared architecture: Infinity-Parser2-Flash, optimized for low-latency inference with a $3.68\times$ throughput gain over Infinity-Parser-7B, and Infinity-Parser2-Pro, engineered for precision-critical settings. Infinity-Parser2-Pro reaches state-of-the-art 87.6% on olmOCR-Bench and 74.3% on ParseBench, surpassing DeepSeek-OCR-2, PaddleOCR-VL-1.5, and MinerU2.5, with strong generalization to charts, chemical formulas, and document VQA.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.07836 [cs.AI] |
| (or arXiv:2607.07836v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.07836 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zuming Huang [view email]
[v1]
Wed, 8 Jul 2026 18:17:21 UTC (22,579 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.