跳转至

文章背景与核心概要

随着智能体AI（Agentic AI）的不断发展，寻找高效改进自主智能体的方法对于加速递归自我改进（RSI）至关重要。提示词优化已成为权重微调的一种极具潜力的替代方案，它不仅能带来可媲美的性能提升，还能在训练和部署阶段大幅降低计算开销。然而，近期的研究往往倾向于采用过于复杂的提示词优化流程。

本文介绍了朴素提示词优化（Naive Prompt Optimization, NPO），这是一种轻量级的单线谱（single-lineage）方法，它结合教师模型和 rollout 反馈对提示词进行迭代微调。研究表明，NPO 能够以更少的 rollout 次数实现与 GEPA 相当甚至更好的性能；当搭配更强大的教师模型时，NPO 的优势进一步扩大，这证明了更强的教师推理能力可以在很大程度上弥补优化器端复杂搜索策略的不足。该研究为简化提示词优化流程提供了重要的理论与实践依据。

Naive Prompt Optimization: Rethinking the Need for Complex Prompt Search

Authors: Yuan Chang, Xiaoqi Chen
Published: August 27, 2026
Primary Subject: Artificial Intelligence (cs.AI)
Secondary Subject: Computation and Language (cs.CL)
arXiv Identifier: arXiv:2608.27266
License: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
license icon

📌 Executive Summary

随着智能体AI的迅猛发展，跨多样化任务高效提升自主智能体能力，对于加速智能体AI中的递归自我改进（RSI）至关重要。提示词优化已成为一种前景广阔的替代方案，它能以较低的优化和部署计算成本，提供与微调模型权重相媲美的性能提升。

然而，近期的研究进展越来越倾向于使用不必要地复杂的提示词优化器。本文引入了朴素提示词优化（Naive Prompt Optimization, NPO），这是一种轻量级的单线谱方法，利用带有 rollout 反馈的教师模型来迭代修改提示词。

该研究的核心结论包括： * 效率高： NPO 以更少的 rollout 次数实现了与 GEPA 等更复杂方法相当甚至更好的性能。 * 教师模型可扩展性： 当与更强的教师模型配对时，NPO 的性能优势会进一步扩大，这表明更强的教师推理能力可以部分替代优化器端的搜索复杂度。 * 任务通用性： 在交互式环境中，NPO 依然能够与 GEPA 保持竞争力，不过某些不太适合提示词优化的任务仍更倾向于使用 GRPO 等方法。 * 可迁移性： 经 NPO 优化的提示词在直接应用于其他学生模型（尤其是同一模型家族内的模型）时，也能产生相似的性能提升。

归根结底，这些研究结果表明，简单、线性的提示词优化完全可以与明显更复杂、更繁琐的搜索流程相媲美。

As agentic AI advances, efficient methods to improve autonomous agents are critical for recursive self-improvement (RSI). Prompt optimization has emerged as a promising alternative to weight fine-tuning, offering comparable performance boosts with lower computational overhead for both training and deployment.

However, recent research increasingly gravitates toward overly complex prompt optimization procedures. This paper introduces Naive Prompt Optimization (NPO), a lightweight, single-lineage method that iteratively refines prompts using a teacher model paired with rollout feedback.

Key takeaways from the study include: * Efficiency: NPO achieves performance comparable to—or better than—more complex methods like GEPA, while requiring fewer rollouts. * Teacher Scalability: The performance advantage of NPO grows when paired with stronger teacher models, demonstrating that superior teacher reasoning can compensate for complex optimizer-side search strategies. * Task Versatility: In interactive environments, NPO remains competitive against GEPA, though certain tasks less suited for prompt optimization still favor methods like GRPO. * Transferability: Prompts optimized via NPO yield similar performance improvements when applied verbatim to other student models, particularly within the same model family.

Ultimately, the findings suggest that simple, linear prompt optimization can effectively rival significantly more intricate search procedures.

📄 Abstract

Efficiently improving autonomous agents across diverse tasks is central to accelerating recursive self-improvement (RSI) in agentic AI, with prompt optimization emerging as a promising approach capable of delivering performance gains comparable to those achieved by fine-tuning model weights, while reducing computational costs in both optimization and serving. However, recent developments increasingly favor unnecessarily complex prompt optimizers. We introduce Naive Prompt Optimization (NPO), a lightweight single-lineage method that iteratively revises prompts using a teacher model with rollout feedback. NPO achieves comparable or better performance than GEPA with fewer rollouts, and its advantage increases with stronger teacher models, suggesting that stronger teacher reasoning can partially substitute for optimizer-side search complexity. In interactive games, NPO remains broadly competitive with GEPA, while GRPO performs better on some tasks less amenable to prompt optimization. We also show that NPO-optimized prompts elicit similar performance improvements when applied verbatim to other student models, especially across models within the same family. Overall, our preliminary results show that simple, linear prompt optimization can rival substantially more sophisticated and complex search procedures.

在各种任务中高效提升自主智能体，是加速智能体AI中递归自我改进（RSI）的核心。提示词优化作为一种极具前景的方法，能够带来与微调模型权重相媲美的性能提升，同时降低优化和部署阶段的计算成本。然而，近期的发展越来越倾向于使用不必要地复杂的提示词优化器。我们引入了朴素提示词优化（NPO），这是一种轻量级的单线谱方法，它利用带有 rollout 反馈的教师模型来迭代修改提示词。NPO 以更少的 rollout 次数实现了与 GEPA 相当或更好的性能，且其优势随着教师模型的增强而扩大，这表明更强的教师推理能力可以部分替代优化器端的搜索复杂度。在交互式游戏中，NPO 保持了与 GEPA 的广泛竞争力，而 GRPO 在某些不太适合提示词优化的任务上表现更好。我们还表明，经 NPO 优化的提示词在直接应用于其他学生模型（尤其是同一模型家族内的模型）时，能引发类似的性能提升。总体而言，我们的初步结果表明，简单、线性的提示词优化完全可以与明显更复杂、更精细的搜索流程相媲美。