Multi-Agent Reinforcement Learning on SMAC Maps benchmark leaderboard · SOTA2 Research

Multi-Agent Reinforcement Learning on SMAC Maps

595m_vs_6m Scoredmix-DA

Evaluation Results

Method	Links
dmix-DA2022.09		59	91.7	46.2	99.8	40.6	64.4	67	—	—	—
dmix2022.09		58.7	76.2	0	99.4	3.3	64.7	50.4	—	—	—
qmix-DA2022.09		50.6	93.3	55.4	99.2	43.9	70.8	68.9	—	—	—
qplex-DA2022.09		38.9	88.3	55.4	99.6	31.6	64.8	63.1	—	—	—
qmix2022.09		36.4	90.2	26.2	98.3	1.9	33.9	47.8	—	—	—
MAPPO2022.05		21.9	—	36.2	100	—	—	—	26.8	100	97.5
MAT2022.05		18.8	—	62.5	100	—	—	—	71.2	100	100
MAT2022.05		5.8	—	33.8	100	—	—	—	53.8	100	82.5
qplex2022.09		5	87	0	98.7	18.9	36.2	41	—	—	—
MAPPO2022.05		4.3	—	13.8	100	—	—	—	26.2	100	73.8
MAT-from scratch2022.05		3.8	—	0	92.5	—	—	—	0.3	96.3	87.5
MAT-from scratch2022.05		1.9	—	0	10.6	—	—	—	0	19.3	7.5
MAT2022.05		0	—	0	100	—	—	—	0	0	3.1
MAT2022.05		0	—	0	100	—	—	—	6.3	15.6	5.6
MAPPO2022.05		0	—	0	100	—	—	—	9.4	0	0
MAPPO2022.05		0	—	0	100	—	—	—	15	43.1	4.3
MAT-from scratch2022.05		0	—	0	0	—	—	—	0	0	0
MAT-from scratch2022.05		0	—	0	0	—	—	—	0	0	0