通过问答形式增强推理中间步骤监督，提升复杂任务准确率。
DAIS: Dependency-Aware Intermediate QA Supervision for Complex Reasoning
- 将教师推理转化为分阶段问答记录，每步依赖前序状态
- 在多个基准上平均提升4.2%准确率，最大增益5.6%
- 轻量级设计，适配主流大模型，无需额外输入
链式思维（CoT）监督暴露中间推理过程，但平坦的推理目标通常只优化单一推理路径，对局部结论如何支撑后续决策监督有限。我们提出依赖感知的中间问答监督（DAIS），一种训练阶段框架，将筛选后的教师推理转换为阶段级问答记录。每个中间记录基于前序状态预测局部答案，最终答案记录保持原任务格式；评估仅需原始输入和可选上下文。在GDPR、AIACT、MedQA和FOLIO多个数据集上，使用多种Qwen骨干模型，DAIS在最终答案准确率上超越仅答案、平坦链式思维和独立问答基线。在政策合规基准上，相比最强非DAIS基线，最大提升5.6%，平均提升4.2%。受控消融实验表明，有效前序状态条件作用超过更长目标或额外中间文本，支持依赖条件中间问答作为标准最终答案推理的轻量级辅助监督信号。
原文摘要 · Abstract (English)
Chain-of-thought (CoT) supervision exposes intermediate rationales, but flat rationale targets usually optimize a single reasoning sequence and provide limited supervision on how local conclusions should support later decisions. We introduce Dependency-Aware Intermediate QA Supervision (DAIS), a training-time framework that converts filtered teacher rationales into stage-level QA records. Each intermediate record predicts a local answer conditioned on the previous states needed for that decision, while the final-answer record keeps the original task format; evaluation therefore uses only the original input and optional context. Across GDPR, AIACT, MedQA, and FOLIO with multiple Qwen backbones, DAIS improves average final-answer accuracy over answer-only, flat chain-of-thought, and independent-QA baselines. On policy-compliance benchmarks, it achieves a largest gain of 5.6% and an average gain of 4.2% over the strongest non-DAIS baseline. Controlled ablations show that valid previous-state conditioning contributes beyond longer targets or additional intermediate text, supporting dependency-conditioned intermediate QA as a lightweight auxiliary supervision signal for standard final-answer inference.
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。