ResearchBenchmarksMulti-Agent Reinforcement Learning on SMAC 8m to 3m (Final Win Rate)Follow96Final Win RateBaseline90.892.1593.594.85Jun 23, 2026Evaluation ResultsMethodMethodLinksFinal Win RateBaselinetraining=from scratch...training=from scratch using MAPPO2026.0696ASALT2026.0694CORALtransfer_scheme=agent-...transfer_scheme=agent-wise transfer2026.0694LA-QTransformer2026.0694DANNtransfer_scheme=agent-...transfer_scheme=agent-wise transfer2026.0692CycleGANtransfer_scheme=agent-...transfer_scheme=agent-wise transfer2026.0691