让大模型像人一样无意识思考，节省47%推理tokens
Efficient Reasoning via Chain of Unconscious Thought
- 用隐藏层思维引导模型内化推理过程
- 减少47.62%的token使用量，准确率不变
- 适合追求高效推理的AI应用开发者
大型推理模型（LRMs）表现优异，但因推理过程冗长而影响效率。无意识思维理论（UTT）指出，复杂问题可通过内化的认知过程更高效解决。受此启发，我们提出链式无意识思维（CoUT）新范式，引导模型模仿人类无意识思维，内化推理过程以提升令牌效率。具体而言，先通过提示使模型在隐层中进行内部思考，再设计一系列轻量级策略，帮助模型减少冗余令牌，同时保持性能。实验表明，模型具备有益的无意识思维能力，能显著提升效率而不牺牲准确率。大量实验证明，CoUT相比传统Chain of Thought（CoT）在保持相近准确率的前提下，将令牌使用量降低47.62%，如图1所示。CoUT代码已开源：https://github.com/Rohan-GRH/CoUT
原文摘要 · Abstract (English)
Large Reasoning Models (LRMs) achieve promising performance but compromise token efficiency due to verbose reasoning processes. Unconscious Thought Theory (UTT) posits that complex problems can be solved more efficiently through internalized cognitive processes. Inspired by UTT, we propose a new reasoning paradigm, termed Chain of Unconscious Thought (CoUT), to improve the token efficiency of LRMs by guiding them to mimic human unconscious thought and internalize reasoning processes. Concretely, we first prompt the model to internalize the reasoning by thinking in the hidden layer. Then, we design a bag of token-efficient strategies to further help models reduce unnecessary tokens yet preserve the performance. Our work reveals that models may possess beneficial unconscious thought, enabling improved efficiency without sacrificing performance. Extensive experiments demonstrate the effectiveness of CoUT. Remarkably, it surpasses CoT by reducing token usage by 47.62% while maintaining comparable accuracy, as shown in Figure 1. The code of CoUT is available at this link: https://github.com/Rohan-GRH/CoUT
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。