1 paper · 1 filter
Seungdong Yoa, Ye Seul Sim, Suhee Yoon +5
Multi-agent debate has shown promise for improving the reasoning of large language models, yet recent theory suggests its benefits are highly regime-dependent. While interaction ca…