阅读还是推理？面向文档OCR的格式解耦强化学习
Reading or Reasoning? Format Decoupled Reinforcement Learning for Document OCR
针对文档OCR中格式敏感文本（公式、表格）输出熵高的问题，提出基于熵的数据过滤和格式解耦奖励的强化学习微调方法，可复用于其他需要结构化输出的视觉语言模型任务。
原文依据 even advanced OCR models exhibit significantly higher entropy in formatted text (e.g., formula, table) compared to plain text
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论通过OCR模型从图像或扫描文档中读取文本一直是研究人员的长期关注点。直观上，文本读取被视为一项直接的感知任务，现有工作主要集中于构建丰富的数据工程以增强SFT能力。在本工作中，我们观察到即使先进的OCR模型在格式化文本（例如公式、表格）上表现出显著更高的熵，通常比纯文本高一个数量级。这些统计模式表明，先进OCR模型在处理格式敏感的文档时面临高输出不确定性，这提示对多样阅读路径进行推理可能提升OCR性能。为此，我们提出了格式解耦强化学习（FD-RL），利用高熵模式进行针对性优化。我们的方法采用基于熵的数据过滤策略来识别格式密集的实例，并采用针对不同格式类型定制的格式解耦奖励，实现格式级验证而非词元级记忆。FD-RL在OmniDocBench上平均得分90.41，在该广受欢迎的基准上展示了端到端模型中极具竞争力的性能。更重要的是，我们对数据、训练、过滤和奖励策略进行了全面的消融研究，彻底验证了其有效性。
英文原文摘要 1413 字符
Reading text from images or scanned documents via OCR models has been a longstanding focus of researchers. Intuitively, text reading is perceived as a straightforward perceptual task, and existing work primarily focuses on constructing enriched data engineering to enhance SFT capabilities. In this work, we observe that even advanced OCR models exhibit significantly higher entropy in formatted text (e.g., formula, table) compared to plain text, often by an order of magnitude. These statistical patterns reveal that advanced OCR models struggle with high output uncertainty when dealing with format sensitive document, suggesting that reasoning over diverse reading pathways may improve OCR performance. To address this, we propose Format Decoupled Reinforcement Learning (FD-RL) , which leverages high-entropy patterns for targeted optimization. Our approach employs entropy-based data filtration strategy to identify format-intensive instances, and adopt format decoupled rewards tailored to different format types, enabling format-level validation rather than token-level memorization. FD-RL achieves an average score of 90.41 on OmniDocBench, demonstrating highly competitive performance among end-to-end models on this highly popular benchmark. More importantly, we conduct comprehensive ablation studies over data, training, filtering, and rewarding strategies, thoroughly validating their effectiveness.