Hint-before-Solving Prompting: Guiding LLMs to Effectively Utilize Encoded Knowledge
This addresses the challenge of improving reasoning accuracy in LLMs for various domains, representing an incremental enhancement over existing prompting methods.
The paper tackles the problem of LLMs inefficiently using encoded knowledge for reasoning by introducing Hint-before-Solving Prompting (HSP), which guides models to generate hints before solutions, resulting in accuracy improvements such as a 9.7-point gain for Llama2-70B-Chat and 64.3% accuracy for a fine-tuned model.
Large Language Models (LLMs) have recently showcased remarkable generalizability in various domains. Despite their extensive knowledge, LLMs still face challenges in efficiently utilizing encoded knowledge to develop accurate and logical reasoning processes. To mitigate this problem, we introduced Hint-before-Solving Prompting (HSP), which guides the model to generate hints (e.g., specific knowledge or key ideas) for solving the problem and then generate solutions containing intermediate reasoning steps. Since HSP is orthogonal to prompting methods (e.g., Chain-of-Thought (CoT)), we applied HSP to CoT, Least-to-Most, Plan-and-Solve, and Standard promptings. The results of extensive experiments on 6 reasoning benchmarks and 4 open-source LLMs demonstrate that HSP can effectively improve the accuracy of reasoning tasks: (1) By applying high-quality hint-enhanced HSP to CoT prompting, Llama2-70B-Chat shows an improvement of 9.7. (2) Beyond exploring training-free LLM capabilities, we built the HSPMATH dataset based on HSP and fine-tuned Llemma-7B, reaching 64.3 accuracy, surpassing GPT-3.5 and WizardMath-13B. We make our code and dataset publicly available at \url{https://github.com/jinlanfu/HSP}.