Adaptive reasoning is essential for aligning the computational effort of large language models (LLMs) with the intrinsic difficulty of problems. Current chain-of-thought methods boost reasoning ability but indiscriminately generate long explanations, leading to evident inefficiency. However, existing reinforcement learning approaches to adaptive thinking remain unstable and heavily reward-dependent. Here we propose \textbf{DART}, a supervised \textbf{D}ifficulty-\textbf{A}daptive \textbf{R}easoning \textbf{T}runcation framework that adjusts thinking length according to problem difficulty. By distilling concise reasoning patterns from stronger models, interpolating them into a continuum of reasoning styles, and curating optimal training data that balances correctness and compactness, DART learns when to ``stop thinking''. Across multiple mathematical benchmarks, experimental results demonstrate its remarkable efficiency while preserving or improving accuracy, achieving a significant 81.2\% reasoning truncation (DeepSeek-R1-Distill-Qwen-7B on GSM8K dataset) with 5.33$\times$ computational acceleration. DART provides a stable and general paradigm for efficient reasoning, advancing the development of adaptive intelligence in LLMs.

翻译：自适应推理对于使大型语言模型（LLMs）的计算投入与问题内在难度相匹配至关重要。现有的思维链方法虽能提升推理能力，但会不加区分地生成长篇解释，导致显著的效率低下。然而，当前基于强化学习的自适应思考方法仍不稳定且严重依赖奖励机制。本文提出\\textbf{DART}，一种监督式的\\textbf{难度自适应推理截断}框架，可根据问题难度调整思考长度。通过从更强模型中蒸馏出简洁的推理模式，将其插值为连续的推理风格谱系，并筛选出平衡正确性与紧凑性的最优训练数据，DART学会了何时“停止思考”。在多个数学基准测试中，实验结果表明其在保持或提升准确率的同时实现了卓越的效率，在GSM8K数据集上（DeepSeek-R1-Distill-Qwen-7B模型）达到81.2%的推理截断率，计算加速比达5.33倍。DART为高效推理提供了稳定且通用的范式，推动了LLMs自适应智能的发展。