让AI像人一样先快读定位再细思，文档理解更准更快
CogDoc: Towards Unified thinking in Documents
- 分两步思考：先快速扫描找重点，再聚焦细节深入分析
- 7B模型在复杂图文文档上超越更大商用模型（如GPT-4o）
- 直接强化学习优于先监督微调再强化学习，避免策略冲突
当前文档推理方法在可扩展性（处理长文档）与准确性（捕捉细粒度多模态细节）之间存在根本矛盾。为此，我们提出CogDoc——一种模仿人类认知过程的统一粗到精思维框架：第一阶段为低分辨率‘快速阅读’，实现长上下文信息定位；第二阶段为高分辨率‘专注思考’，支持深度推理。我们系统研究了该统一框架的后训练策略，发现直接强化学习（Direct RL）优于以监督微调（SFT）初始化的强化学习，因后者存在‘策略冲突’问题。实验证明，我们的7B模型在同参数量级中达到领先性能，尤其在视觉丰富的复杂文档基准测试中显著超越更大规模的闭源模型（如GPT-4o）。
原文摘要 · Abstract (English)
Current document reasoning paradigms are constrained by a fundamental trade-off between scalability (processing long-context documents) and fidelity (capturing fine-grained, multimodal details). To bridge this gap, we propose CogDoc, a unified coarse-to-fine thinking framework that mimics human cognitive processes: a low-resolution "Fast Reading" phase for scalable information localization,followed by a high-resolution "Focused Thinking" phase for deep reasoning. We conduct a rigorous investigation into post-training strategies for the unified thinking framework, demonstrating that a Direct Reinforcement Learning (RL) approach outperforms RL with Supervised Fine-Tuning (SFT) initialization. Specifically, we find that direct RL avoids the "policy conflict" observed in SFT. Empirically, our 7B model achieves state-of-the-art performance within its parameter class, notably surpassing significantly larger proprietary models (e.g., GPT-4o) on challenging, visually rich document benchmarks.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。