# MA-SAPO: Multi-Agent Reasoning for Score-Aware Prompt Optimization

Source: [MA-SAPO: Multi-Agent Reasoning for Score-Aware Prompt Optimization](<https://www.sota2.com/research/paper/68fa7c551e209131cea8d9ae>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-02T05:00:04.464Z

Authors: Wonduk Seo, Juhyeon Lee, Junseo Koh, Wonseok Choi, Hyunjin An, Jian Park, Seunghyun lee, Haihua Chen, Yi Bu

Published: 2025-10-18T20:21:09.000Z

## Abstract

Prompt optimization has become a practical way to improve the performance of Large Language Models (LLMs) without retraining. However, most existing frameworks treat evaluation as a black box, relying solely on outcome scores without explaining why prompts succeed or fail. Moreover, they involve repetitive trial-and-error refinements that remain implicit, offering limited interpretability or actionable guidance for systematic improvement. In this paper, we propose MA-SAPO: a new Multi-Agent Reasoning for Score Aware Prompt Optimization framework that links evaluation outcomes directly to targeted refinements. Specifically, in the Training Phase, multiple agents interpret evaluation scores, diagnose weaknesses, and generate concrete revision directives, which are stored as reusable reasoning assets. In the Test Phase, an analyzer agent retrieves relevant exemplars and assets for a new prompt, and a refiner agent applies evidence-based edits to improve the prompt and its response. By grounding optimization in structured reasoning, MA-SAPO ensures edits are interpretable, auditable, and controllable. Experiments on the HelpSteer1/2 benchmarks show that our framework consistently outperforms single-pass prompting, retrieval-augmented generation, and prior multi-agent methods across multiple evaluation metrics.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Prompt Optimization Evaluation on HelpSteer 1](<https://www.sota2.com/research/sota/prompt-optimization-evaluation-on-helpsteer-1.md>): Prompt Optimization Evaluation; HelpSteer 1; Helpfulness: 51.83.

- [Prompt Optimization Evaluation on HelpSteer2](<https://www.sota2.com/research/sota/prompt-optimization-evaluation-on-helpsteer2.md>): Prompt Optimization Evaluation; HelpSteer2; Helpfulness: 0.5072.

- [Reasoning Quality Evaluation on HelpSteer1 sampled (train)](<https://www.sota2.com/research/sota/reasoning-quality-evaluation-on-helpsteer1-sampled-train.md>): Reasoning Quality Evaluation; HelpSteer1 sampled (train); Usefulness Score: 3.89.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2510.16635>)

[PDF](<https://arxiv.org/pdf/2510.16635.pdf>)