---
id: 20260519-T0-09
title: "TeamTR：多智能体LLM协同训练的新方案"
title_en: "TeamTR: Trust-Region Tuning for Multi-Agent LLMs"
url: https://ai.daily.yangsir.net/daily/20260519-T0-09
issue_date: 2026-05-19
publish_date: 2026-05-18T04:00:00.000Z
category: research
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2605.15207
---

# TeamTR：多智能体LLM协同训练的新方案

arXiv论文TeamTR解决多智能体LLM系统协同问题。研究发现传统顺序微调会导致智能体间累积误差，新方法采用信任域微调提升性能。在复杂推理任务中，TeamTR比基线模型准确率提高23%，智能体协作效率显著改善。

## English Version

**TeamTR: Trust-Region Tuning for Multi-Agent LLMs**

TeamTR paper identifies sequential fine-tuning issues in multi-agent LLM systems and proposes trust-region tuning to improve coordination. It achieves 23% higher accuracy in complex reasoning tasks compared to baselines.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2605.15207)

**详情页**：https://ai.daily.yangsir.net/daily/20260519-T0-09

---

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