Multi Agent Framework
Multi-agent frameworks leverage the collaborative power of multiple artificial intelligence agents, often based on large language models (LLMs), to solve complex problems exceeding the capabilities of individual agents. Current research emphasizes efficient communication protocols to reduce computational costs and improve robustness against adversarial attacks, as well as the development of specialized agents for diverse tasks, including code generation, patent analysis, and urban mobility management. These frameworks are proving valuable for automating intricate workflows across various domains, improving efficiency and accuracy in areas like software development, financial analysis, and data reporting.
Papers
Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence
Weize Chen, Ziming You, Ran Li, Yitong Guan, Chen Qian, Chenyang Zhao, Cheng Yang, Ruobing Xie, Zhiyuan Liu, Maosong Sun
PEER: Expertizing Domain-Specific Tasks with a Multi-Agent Framework and Tuning Methods
Yiying Wang, Xiaojing Li, Binzhu Wang, Yueyang Zhou, Yingru Lin, Han Ji, Hong Chen, Jinshi Zhang, Fei Yu, Zewei Zhao, Song Jin, Renji Gong, Wanqing Xu