让小模型学会大模型的空间推理，精准定位文档答案。
DocVAL: Validated Chain-of-Thought Distillation for Grounded Document VQA
- 用大模型生成带空间逻辑的思考链，指导小模型学习
- 通过验证器过滤差数据，提升定位准确率最高达6-7 ANLS点
- 无需文字识别就能推理，适合实际部署的轻量级文档问答
文档视觉问答要求模型不仅正确回答问题，还需在复杂文档布局中精确定位答案。尽管大模型具备强大的空间定位能力，但其推理成本和延迟限制了实际应用。紧凑型模型虽更高效，但在常规微调或蒸馏后常出现定位性能显著下降。为此，我们提出DocVAL，一种基于验证链式思维（CoT）的蒸馏框架，将大模型的显式空间推理知识迁移至可部署的小型学生模型。DocVAL结合三项机制：(1) 教师模型生成的空间推理链监督；(2) 基于规则的双模式验证器，用于过滤低质量训练信号并提供像素级修正反馈；(3) 验证驱动的两阶段迭代训练流程。文本检测仅作为训练期辅助工具，学生模型最终可在无OCR或检测的纯视觉语言模型下运行。在多个文档理解基准上，DocVAL相较同类紧凑模型实现最高6–7 ANLS点的持续提升。我们进一步引入平均精度（mAP）作为文档问答的定位评估指标，并报告该标准下的强空间定位性能。我们公开95K条经验证器确认的CoT轨迹，表明高质量、经验证的监督比单纯扩大未过滤数据更有效，支持高效且可信的文档定位。代码与数据见：https://github.com/ahmad-shirazi/DocVAL
原文摘要 · Abstract (English)
Document visual question answering requires models not only to answer questions correctly, but also to precisely localize answers within complex document layouts. While large vision-language models (VLMs) achieve strong spatial grounding, their inference cost and latency limit real-world deployment. Compact VLMs are more efficient, but they often suffer substantial localization degradation under standard fine-tuning or distillation. To address this gap, we propose DocVAL, a validated chain-of-thought (CoT) distillation framework that transfers explicit spatial reasoning from large teacher models to compact, deployable student VLMs. DocVAL combines (1) teacher-generated spatial CoT supervision, (2) a rule-based dual-mode validator that filters low-quality training signals and provides fine-grained, pixel-level corrective feedback, and (3) a validation-driven two-stage training procedure with iterative refinement. Text detection is used only as training-time scaffolding for supervision and validation, enabling the final student to operate as a pure VLM without OCR or detection at inference. Across multiple document understanding benchmarks, DocVAL yields consistent improvements of up to 6-7 ANLS points over comparable compact VLMs. We further introduce mean Average Precision (mAP) as a localization metric for document question answering and report strong spatial grounding performance under this new evaluation. We release 95K validator-verified CoT traces and show that high-quality, validated supervision is more effective than scaling unfiltered data, enabling efficient and trustworthy document grounding. Code/Data: https://github.com/ahmad-shirazi/DocVAL
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。