List your company

Multi-Agent Reinforcement Learning on SMAC v2 (test)

84Win Rate (Protoss 5 Units)HPN-QMIX
-3.349619.327742.00564.6823Sep 25, 2025Nov 3, 2025Dec 12, 2025Jan 20, 2026Feb 28, 2026Apr 8, 2026May 18, 2026

Evaluation results

MethodLinks
HPN-QMIX2026.02848278—
HPN-VDN2026.02847079—
MMSA2026.02818381—
QPLEX-ICES2025.0974—48—
QMIX-CERMIC2025.0973—44—
MAPPO-CERMIC2025.0970—48—
QMIX2026.02697573—
MACE2025.0967—44—
QMIX-SPECTRA2025.0965—45—
CPM2025.0964—44—
SET-QMIX2026.02636464—
MAPPO-DB2025.0963—43—
MAPPO2025.0961—44—
MAPPOexploration=epsilon greedy2025.0961—43—
QMIX2025.0960—39—
MAG2026.02585934—
LMACEnvironment steps=3M2026.0557.9667.8742.18—
QMIX+STATEEnvironment steps=3M2026.0556.464.7740.06—
MAMBA2026.02545432—
MAICEnvironment steps=3M2026.0551.9363.838.59—
QMIXEnvironment steps=3M2026.0548.4461.6934.79—
T2MACEnvironment steps=3M2026.0548.1661.6735.09—
NDQEnvironment steps=3M2026.0548.0359.238.75—
VDN2026.02466762—
MAPPO2026.02384529—
STAIRS-Former2026.0332.832.32530
MASIAEnvironment steps=3M2026.0532.4354.7234.22—
HiSSD2026.0328.525.118.824.1
SMSEnvironment steps=3M2026.05273413.53—
TarMACEnvironment steps=3M2026.0522.5629.6815.47—
FullCommEnvironment steps=3M2026.0516.9318.446.83—
ODIS2026.0311.613.2810.9
MADDPG2026.0210209—
UPDeT-m2026.038.1756.7
COLAEnvironment steps=3M2026.050.0110.290.19—
DGNBaseline Architecture=DGN2026.06———17.48
KNIFEBaseline Architecture=MAPPO2026.06———25.44
KNIFEBaseline Architecture=MADDPG2026.06———34.55
KNIFEBaseline Architecture=DGN2026.06———22.73
MADDPGBaseline Architecture=MADDPG2026.06———20.13
MAPPOBaseline Architecture=MAPPO2026.06———18.84
ReBornBaseline Architecture=MAPPO2026.06———17.61
ReBornBaseline Architecture=MADDPG2026.06———29.16
ReDoBaseline Architecture=MAPPO2026.06———20.26
ReDoBaseline Architecture=MADDPG2026.06———10.48