面向复杂任务异构大模型智能体协同决策OA

[Research purpose]Multi-agent collaboration has been an important direction for enhancing the complex task processing capabilityof large models in recent years.Under the premise that multiple different models constitute agents and carry out collaboration,more efficient decision-makingcan be achieved relying on task division and information exchange.[Research method]This paper explores this method from the perspective of agent collaboration of large models for complex tasks,analyzes it from the dimensions of heterogeneous agent architecture,multi-agent collaborative decision-making mechanism,and complex task matching,and designs a collaborative decision-making and task scheduling method for heterogeneous large model agents for complex tasks on this basis.[Research conclusion]In the above architecture,models interact in a unified way and dynamically allocate different tasks according to the current situation;at the same time,they can adjust the information mastered by themselves according to requirements,which to a certain extent improves the robustness and intelligence of the system.Therefore,this paper proposes a communication and task scheduling method for large model agents oriented to complex tasks.