Large Language Models Miss the Multi-agent Mark

Neural Information Processing Systems 

Recent interest in Multi-Agent Systems of Large Language Models (MAS LLMs) has led to an increase in frameworks leveraging multiple LLMs to tackle complex tasks.