---
id: 20260815-T0-10
title: "SAPO提示词优化新方法：按角色/上下文/任务分段调整，不互相拖累"
title_en: "SAPO: Segment-Level Prompt Optimization Improves One Behavior Without Hurting Others"
url: https://ai.daily.yangsir.net/daily/20260815-T0-10
issue_date: 2026-08-15
publish_date: 2026-08-14T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2608.11219
---

# SAPO提示词优化新方法：按角色/上下文/任务分段调整，不互相拖累

传统的自动提示词优化通常整体重写整段prompt，导致改好一个行为的同时破坏其他行为。SAPO将prompt拆分为角色、上下文、任务和输出格式四个段落，针对问题逐段优化，精准改动而不影响其他部分。实验显示SAPO在多个指令跟随基准上全面优于整体优化方法，并能在优化后保持原prompt中未出问题的行为不变。

## English Version

**SAPO: Segment-Level Prompt Optimization Improves One Behavior Without Hurting Others**

Automatic prompt optimization often rewrites entire prompts monolithically, fixing one behavior while breaking others. SAPO decomposes prompts into role, context, task, and output format segments, optimizing only the problematic segment. It outperforms monolithic optimization on instruction-following benchmarks while preserving previously correct behaviors, offering a precise alternative for prompt tuning.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2608.11219)

**详情页**：https://ai.daily.yangsir.net/daily/20260815-T0-10

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*