ResearchBenchmarksMulti-Agent Reinforcement Learning Transfer on SMAC 8m to 8mvs9mMulti-Agent Reinforcement Learning Transfer on SMAC 8m to 8mvs9mFollow0.53Steps to 80% Reward (x10^6)GCT-MARL0.31761.75133.1854.6187Jun 23, 2026Evaluation resultsMethodMethodLinksSteps to 80% Reward (x10^6)Final Win RateGCT-MARLScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.060.53100MAILScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.062.25100MALTScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.062.4189EPCScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.063.0289DistilledScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.064.2886PSMARLScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.065.2391BaselineScenario=8m → 8mvs9mScenario=8m → 8mvs9m2026.065.8494