收录解读
TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems 关注的是一个可复用的 AI 系统或评测问题，而不是单点 demo。
Test-time co-evolution of multi-agent topology and capabilities.
It frames MAS inference as online graph adaptation with distinct time scales for capability and topology updates.
它没有更高，是因为这些新 arXiv 工作仍需要更多独立复现、真实系统部署和长期社区采用来确认影响。
原始摘要与中文对照
中文对照翻译
多智能体系统（MAS）已成为解决复杂任务的一种有前景的范式。最近的研究探索了自动优化智能体能力或通信拓扑的自演化MAS。然而，现有方法要么学习在推理时保持固定的拓扑，要么在推理期间仅适应拓扑或能力。我们通过经验和理论证明，有效的测试时演化需要同时适应这两个维度，但要在不同的时间尺度上：能力应快速更新以处理新出现的子任务，而拓扑应更慢地演化以保持协调稳定性。然后，我们引入了TacoMAS，一个用于动态MAS的测试时协同演化框架。TacoMAS将MAS推理表述为在线图适应任务，其中节点代表具有特定角色能力的智能体，边定义其通信拓扑。在推理过程中，一个快速能力循环使用轨迹级反馈更新智能体专业知识，而一个慢速的元LLM驱动拓扑循环在MAS上执行智能体的生灭操作，包括边编辑、智能体添加和智能体移除。我们进一步表明，这种快慢设计将MAS演化推向任务条件下的稳定均衡。在四个基准测试上的实验表明，TacoMAS超越了近20个多智能体基线，比最强的基线平均提高了13.3%。
原始摘要
Multi-agent systems (MAS) have emerged as a promising paradigm for solving complex tasks. Recent work has explored self-evolving MAS that automatically optimize agent capabilities or communication topologies. However, existing methods either learn a topology that remains fixed at inference time or adapt only topology or capability during inference. We empirically and theoretically show that effective test-time evolution requires jointly adapting both axes, but on different time scales: capabilities should update rapidly to handle emerging subtasks, while the topology should evolve more slowly to preserve coordination stability. We then introduce TacoMAS, a test-time co-evolution framework for dynamic MAS. TacoMAS formulates MAS inference as a task of online graph adaptation, where nodes represent agents with role-specific capability and edges define their communication topology. During inference, a fast capability loop updates agent expertise using trajectory-level feedback, while a slow meta-LLM-driven topology loop performs agents’ birth-death operations on MAS, including edge edit, agent addition, and agent removal. We further show that this fast–slow design drives MAS evolution toward a task-conditioned stable equilibrium. Experiments on four benchmarks demonstrate that TacoMAS outperforms nearly 20 multi-agent baselines, achieving an average improvement of 13.3% over the strongest baseline.