Prepare Reasoning Language Models for Multi-Agent Debate with Self-Debate Reinforcement Learning

LLMAGReLMLRMRe-assign community