HPSS: Heuristic Prompting Strategy Search for LLM Evaluators

TL;DR

This paper introduces HPSS, an automatic heuristic search method inspired by genetic algorithms, to optimize multiple factors in LLM evaluation prompts, significantly improving alignment with human judgment across various tasks.

Contribution

It presents a novel integrated approach for optimizing multiple prompt factors simultaneously, addressing limitations of prior methods that focus on individual factors.

Findings

HPSS outperforms human-designed prompts in multiple evaluation tasks.

The heuristic search effectively finds well-behaved prompting strategies.

Extensive experiments validate the method's superiority over existing automatic optimization techniques.

Abstract

Since the adoption of large language models (LLMs) for text evaluation has become increasingly prevalent in the field of natural language processing (NLP), a series of existing works attempt to optimize the prompts for LLM evaluators to improve their alignment with human judgment. However, their efforts are limited to optimizing individual factors of evaluation prompts, such as evaluation criteria or output formats, neglecting the combinatorial impact of multiple factors, which leads to insufficient optimization of the evaluation pipeline. Nevertheless, identifying well-behaved prompting strategies for adjusting multiple factors requires extensive enumeration. To this end, we comprehensively integrate 8 key factors for evaluation prompts and propose a novel automatic prompting strategy optimization method called Heuristic Prompting Strategy Search (HPSS). Inspired by the genetic algorithm, HPSS conducts an iterative search to find well-behaved prompting strategies for LLM evaluators. A heuristic function is employed to guide the search process, enhancing the performance of our algorithm. Extensive experiments across four evaluation tasks demonstrate the effectiveness of HPSS, consistently outperforming both human-designed evaluation prompts and existing automatic prompt optimization methods. Our code is available at https://github.com/thu-coai/HPSS.