---
id: 20260318-T0-25
title: "REDEREF：训练多智能体LLM系统的新方法"
title_en: "REDEREF: New Method for Training Multi-Agent LLM Systems"
url: https://ai.daily.yangsir.net/daily/20260318-T0-25
issue_date: 2026-03-18
publish_date: 2026-03-17T04:00:00.000Z
source_name: "arXiv cs.CL (NLP)"
source_url: https://arxiv.org/abs/2603.13256
---

# REDEREF：训练多智能体LLM系统的新方法

研究人员提出REDEREF框架，解决多智能体LLM系统的路由效率低、反馈噪声高、交互成本大等问题。通过概率控制和协调机制，使多个专业代理能高效协作完成复杂任务。实验显示在多步骤推理任务中比基线模型提升32%成功率。

## English Version

**REDEREF: New Method for Training Multi-Agent LLM Systems**

Researchers introduced REDEREF, addressing inefficiencies in multi-agent LLM systems including poor routing, noisy feedback, and high interaction costs. Using probabilistic control and coordination, it enables specialized agents to collaborate efficiently on complex tasks. Experiments show 32% higher success rates on multi-step reasoning tasks.

---

**来源**：[arXiv cs.CL (NLP)](https://arxiv.org/abs/2603.13256)

**详情页**：https://ai.daily.yangsir.net/daily/20260318-T0-25

---

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