---
id: 20260522-T0-12
title: "系统提示优化新方法：基于提示的动态表示学习"
title_en: "Dynamic Prompt Embeddings for System Prompt Optimization"
url: https://ai.daily.yangsir.net/daily/20260522-T0-12
issue_date: 2026-05-22
publish_date: 2026-05-21T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2605.19093
---

# 系统提示优化新方法：基于提示的动态表示学习

arXiv论文提出通过提示嵌入优化AI系统提示的新方法。传统系统提示优化依赖聚合指标反馈，难以逐例调整。该研究设计动态表示机制，将用户反馈转化为可计算的系统提示向量，实现更精准的参数调整。实验显示该方法在对话任务中减少22%的无效响应，尤其适用于需要个性化响应的客服场景。

## English Version

**Dynamic Prompt Embeddings for System Prompt Optimization**

arXiv paper introduces dynamic prompt embeddings to optimize AI system prompts. Traditional tuning struggles with aggregate feedback metrics. This method transforms user feedback into computable system prompt vectors, enabling precise parameter adjustments. Experiments show 22% reduction in invalid responses for dialogue tasks, particularly valuable for customer service requiring personalized interactions.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2605.19093)

**详情页**：https://ai.daily.yangsir.net/daily/20260522-T0-12

---

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