Multi-agent reinforcement learning on SMAC 5v5 full 3x5 grid v2
81Win Rate (Protoss)FULLOBS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FULLOBSAlgorithm=QMIX2026.06 | 81 | 84 | 63 | |
| HiCommAlgorithm=QMIX2026.06 | 78 | 81 | 56 | |
| FULLOBSAlgorithm=MAPPO2026.06 | 78 | 72 | 63 | |
| FULLOBSAlgorithm=IPPO2026.06 | 75 | 78 | 59 | |
| HiCommAlgorithm=MAPPO2026.06 | 72 | 66 | 56 | |
| PARTIALOBSAlgorithm=QMIX2026.06 | 69 | 69 | 44 | |
| HiCommAlgorithm=IPPO2026.06 | 69 | 72 | 53 | |
| CACOMAlgorithm=MAPPO2026.06 | 66 | 59 | 50 | |
| CACOMAlgorithm=IPPO2026.06 | 63 | 66 | 44 | |
| T2MACAlgorithm=MAPPO2026.06 | 63 | 56 | 47 | |
| T2MACAlgorithm=IPPO2026.06 | 59 | 63 | 44 | |
| PARTIALOBSAlgorithm=MAPPO2026.06 | 56 | 50 | 44 | |
| CACOMAlgorithm=QMIX2026.06 | 53 | 69 | 46 | |
| PARTIALOBSAlgorithm=IPPO2026.06 | 53 | 56 | 38 | |
| T2MACAlgorithm=QMIX2026.06 | 50 | 66 | 44 |