Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wang, Zheng, Liu, Yuang, Ding, Yangkai
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2605.12943
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911679320686592
author Wang, Zheng
Liu, Yuang
Ding, Yangkai
author_facet Wang, Zheng
Liu, Yuang
Ding, Yangkai
contents Multi-agent systems provide a powerful way to extend large language models (LLMs) by decomposing a complex task into specialized subtasks handled by different agents. However, their performance is often hindered by error propagation, arising from suboptimal workflow design or inaccurate agent outputs, which can propagate through the agent collaboration process and degrade final results. To address the challenges, we present MANGO (Multi-Agent Network Gradient Optimization), a data-driven framework that organizes and refines agent collaboration via a flow network constructed from past successful workflows. MANGO integrates reinforcement learning and textual gradients to jointly optimize workflow paths and agent behaviors, while a skipping mechanism prevents redundant updates to well-optimized agents for improving efficiency. Extensive experiments on seven benchmarks show that MANGO achieves up to 12.8% performance improvement over state-of-the-art baselines, enhances efficiency by 47.4%, and generalizes effectively to unseen domains. Our code and datasets are publicly available at https://github.com/openJiuwen-ai/agent-store/tree/main/community/mango.
format Preprint
id arxiv_https___arxiv_org_abs_2605_12943
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Reinforced Collaboration in Multi-Agent Flow Networks
Wang, Zheng
Liu, Yuang
Ding, Yangkai
Machine Learning
Multi-agent systems provide a powerful way to extend large language models (LLMs) by decomposing a complex task into specialized subtasks handled by different agents. However, their performance is often hindered by error propagation, arising from suboptimal workflow design or inaccurate agent outputs, which can propagate through the agent collaboration process and degrade final results. To address the challenges, we present MANGO (Multi-Agent Network Gradient Optimization), a data-driven framework that organizes and refines agent collaboration via a flow network constructed from past successful workflows. MANGO integrates reinforcement learning and textual gradients to jointly optimize workflow paths and agent behaviors, while a skipping mechanism prevents redundant updates to well-optimized agents for improving efficiency. Extensive experiments on seven benchmarks show that MANGO achieves up to 12.8% performance improvement over state-of-the-art baselines, enhances efficiency by 47.4%, and generalizes effectively to unseen domains. Our code and datasets are publicly available at https://github.com/openJiuwen-ai/agent-store/tree/main/community/mango.
title Reinforced Collaboration in Multi-Agent Flow Networks
topic Machine Learning
url https://arxiv.org/abs/2605.12943