Debate Only When Necessary: Adaptive Multiagent Collaboration for Efficient LLM Reasoning

Open in new window