ResearchBenchmarksMulti-Agent Reinforcement Learning on sce3m (train)Multi-Agent Reinforcement Learning on sce3m (train)Follow100Win RateMAPPO89.39292.14694.997.654Jun 29, 2026Evaluation resultsMethodMethodLinksWin RateFinal ScoreMAPPO2026.0610075.329QTRAN2026.0698.345.814COMA2026.0698.257.033QMIX2026.0697.942.576IQL2026.069740.839VDN2026.0695.942.816HRL-IM/CBS2026.0689.844.794