Beyond Answer Sharing: TrueMAD for Targeted Peer Engagement in Multi-Agent Debate
Multi-agent debate (MAD) aims to improve large language model reasoning by exposing agents to alternative solutions and enabling iterative revision. However, standard broadcast-style MAD makes peer answers visible without requiring agents to engage with a particular peer claim. As a result, reliable reasoning may be su...
De-Zheng Bao, Si-Hang Du, Ren-Hong Huang et al.
· 0 citations