Benefits and Limitations of Communication in Multi-Agent Reasoning

Michael Rizvi-Martel, Satwik Bhattamishra, Neil Rathi +2

Chain-of-thought prompting has popularized step-by-step reasoning in large language models, yet model performance still degrades as problem complexity and context length grow. By d…