Offline Multi-Agent Reinforcement Learning on SMAC Marine-Hard Medium benchmark leaderboard · SOTA2 Research

Offline Multi-Agent Reinforcement Learning on SMAC Marine-Hard Medium

84.4Win Rate (3m)STAIRS

Evaluation Results

Method	Links
STAIRS2026.03		84.4	50	86.9	89.4	100	97.5	95.6	10.6	15.6	61.3	1.3	1.9	57.9
HiSSD2026.03		65	35.6	68.1	78.1	94.4	96.3	88.1	5.6	14.4	46.3	1.3	0.6	49.5
ODIS2026.03		59.4	22.5	57.5	71.3	80.6	65.6	56.3	6.3	10.6	19.4	0	0	37.5
UPDeT-m2026.03		33.8	1.9	31.3	27.5	59.4	49.4	33.1	1.3	1.3	4.4	0	0	20.3