The performance of Large Language Models (LLMs) in reasoning tasks depends heavily on prompt design, with Chain-of-Thought (CoT) and self-consistency being critical methods that enhance this ability. However, these methods do not fully exploit the answers generated by the LLM to guide subsequent responses. This paper proposes a new prompting method, named Progressive-Hint Prompting (PHP), that enables automatic multiple interactions between users and LLMs by using previously generated answers as hints to progressively guide toward the correct answers. PHP is orthogonal to CoT and self-consistency, making it easy to combine with state-of-the-art techniques to further improve performance. We conducted an extensive and comprehensive evaluation to demonstrate the effectiveness of the proposed method. Our experimental results on seven benchmarks show that combining CoT and self-consistency with PHP significantly improves accuracy while remaining highly efficient. For instance, with text-davinci-003, we observed a 4.2% improvement on GSM8K with greedy decoding compared to Complex CoT, and a 46.17% reduction in sample paths with self-consistency. With GPT-4 and PHP, we achieve state-of-the-art performances on SVAMP (89.1% -> 91.9%), GSM8K (92% -> 95.5%), AQuA (76.4% -> 79.9%) and MATH (50.3% -> 53.9%).

翻译：大型语言模型（LLMs）在推理任务中的性能高度依赖于提示设计，其中思维链（CoT）和自一致性是增强这一能力的关键方法。然而，这些方法并未充分利用LLM生成的答案来指导后续回应。本文提出了一种名为渐进提示（Progressive-Hint Prompting, PHP）的新提示方法，通过将先前生成的答案作为提示，逐步引导用户与LLM之间自动进行多次交互，从而趋近正确答案。PHP与CoT和自一致性正交，便于与现有最优技术结合以进一步提升性能。我们进行了广泛而全面的评估以证明该方法的有效性。在七个基准测试上的实验结果表明，将CoT和自一致性与PHP结合，能在保持高效的同时显著提升准确性。例如，使用text-davinci-003时，在GSM8K上贪心解码相比Complex CoT提升了4.2%的准确率，而在自一致性方法中样本路径减少了46.17%。结合GPT-4与PHP，我们在SVAMP（89.1% -> 91.9%）、GSM8K（92% -> 95.5%）、AQuA（76.4% -> 79.9%）和MATH（50.3% -> 53.9%）上达到了当前最优性能。