CogDoc: Towards Unified thinking in Documents
This addresses the problem of scalable and high-fidelity document reasoning for AI applications, representing a novel method rather than an incremental improvement.
The paper tackles the trade-off between scalability and fidelity in document reasoning by proposing CogDoc, a unified coarse-to-fine thinking framework that mimics human cognitive processes, and their 7B model achieves state-of-the-art performance, surpassing larger models like GPT-4o on visually rich document benchmarks.
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.