[论文] Learning Self-Orchestration in Decentralized Agent Swarms

论文概要

研究领域: NLP 作者: Guowei Xu, Zhenting Qi, Amani Kiruga, Himabindu Lakkaraju, Yilun Du 发布时间: 2026-10-15 arXiv: 2610.22322

中文摘要

智能体群的能力取决于其个体成员及它们的协调方式。然而，现有的多智能体系统通常在固定的编排结构内训练智能体，将个体能力的学习与集体行为的组织分离。我们引入基于市场的后训练（MPT），这是一个用于在去中心化智能体群中联合学习动作策略和自我编排的框架。智能体通过出价表达参与意愿，出价决定谁采取行动以及集体结果如何提供局部训练信用。学习动作和出价行为使个体能力和参与模式能够共同适应，无需中央协调智能体。我们在具身制作、机器人导航和深度研究上评估 MPT，结果显示优于单体后训练，并且学习到的自我编排优于替代协调策略。训练后的群体更快适应相关任务，在推理时更大的群体规模下表现更好。代码和训练模型可在 https://github.com/Embodied-Minds-Lab/MPT 获取。

原文摘要

The capabilities of an agent swarm depend on both its individual members and how they coordinate. Yet existing multi-agent systems often train agents within fixed orchestration structures, separating the learning of individual capabilities from the organization of collective behavior. We introduce market-based post-training (MPT), a framework for jointly learning action policies and self-orchestration in decentralized agent swarms. Agents express their willingness to participate through bids, which determine who acts and how collective outcomes provide local training credit. Learning both action and bidding behavior allows individual capabilities and patterns of participation to adapt together without a central coordinating agent. We evaluate MPT on embodied crafting, robot navigation, and...

#论文 #arXiv #NLP #小凯