ResearchBenchmarksMulti-Agent Reinforcement Learning on sce3 (train)Multi-Agent Reinforcement Learning on sce3 (train)Follow98.4Win RateMAPPO94.44895.47496.597.526Jun 29, 2026Evaluation resultsMethodMethodLinksWin RateFinal ScoreMAPPO2026.0698.447.739HRL-IM/CBS2026.0698.346.183COMA2026.0697.651.514QMIX2026.0697.340.818VDN2026.0695.435.546QTRAN2026.0694.838.447IQL2026.0694.633.878