用激活差异聚类生成推理原型，动态引导大模型思考。
Prototype-Based Dynamic Steering for Large Language Models
- 通过聚类思维链与中性提示的激活差异，构建推理原型。
- 在GSM8K等任务上提升准确率，无需微调或提示工程。
- 适用于希望低成本增强模型推理能力的研究者。
尽管表现广泛，大语言模型仍依赖显式推理指令或静态统一的引导方法，难以实现自适应、无指令的推理增强。本文提出原型驱动的动态引导（PDS），一种测试时方法，可在不添加或修改指令的情况下增强大语言模型的推理能力。通过聚类思维链（CoT）与中性提示之间的激活差异，构建“推理原型”。推理时，将输入的隐藏状态投影到这些原型上，生成实例相关的引导向量。在GSM8K、AQuA-RAT和BIG-Bench任务上评估，PDS持续提升准确率，且无需微调或提示工程。值得注意的是，即使显式抑制思维链以提升成本效率，性能增益依然存在，表明该干预强化了潜在的推理过程，而非诱导表面行为变化。结果表明，动态原型引导是一种轻量级替代训练时方法的推理增强方案。
原文摘要 · Abstract (English)
Despite impressive breadth, LLMs still rely on explicit reasoning instructions or static, one-fits-all steering methods, leaving a gap for adaptive, instruction-free reasoning amplification. We present Prototype-Based Dynamic Steering (PDS), a test-time method that amplifies large language model (LLM) reasoning without adding or altering instructions. We introduce "reasoning prototypes" by clustering activation differences between Chain-of-Thought (CoT) and neutral prompts. At inference, an input's hidden state is projected onto these prototypes to form an instance-specific steering vector. Evaluated on GSM8K, AQuA-RAT, and BIG-Bench tasks, PDS consistently improves accuracy without fine-tuning or prompt engineering. Notably, the gains persist even when CoT is explicitly suppressed to improve cost-efficiency, indicating that the intervention strengthens latent reasoning processes rather than inducing a superficial behavioral shift. These results position dynamic, prototype-guided steering as a lightweight alternative to training-time approaches for enhancing LLM reasoning.
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。