ResearchBenchmarksMulti-agent coordination on CoordSumMulti-agent coordination on CoordSumFollow156.37Reward (3 Agents, 10 Steps, 30 Goal)MAPPO64.766888.5484112.33136.1116Jul 24, 2025Evaluation resultsMethodMethodLinksReward (3 Agents, 10 Steps, 30 Goal)Mean RewardReward (3 Agents, 30 Steps, 50 Goal)Reward (5 Agents, 20 Steps, 80 Goal)Reward (8 Agents, 15 Steps, 100 Goal)MAPPO2025.07156.37—158.05142.51127.32SABLE2025.07153.7—125.0448.3528.91HAPPO2025.07153.32—129.35119.1977.6MAGPO2025.07153.13—156.62157.61129.94CTDS2025.07111.98—76.2719.6223.96MAT2025.0768.29—87.7986.8657.22