ResearchBenchmarksCollaborative Multi-Agent Reinforcement Learning on SMAC Zerg (5 agents) v2Follow49.1Win RateGRASP35.78839.24442.746.156Apr 1, 2026Evaluation ResultsMethodMethodLinksWin RateGRASPbase_method=MAPPObase_method=MAPPO2026.0449.1MAPPO2026.0441.8MA2E2026.0439.6HAPPO2026.0436.3