Multi-Agent Coordination on XOR game n=k=2 (train eval)
100Success Rate (Greedy Policy pi*)MAT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MAT2026.05 | 100 | 100 | |
| Diamond Attentionstructured random masking=enabled2026.05 | 100 | 100 | |
| MAPPO2026.05 | 0 | 50 | |
| QMIX2026.05 | 0 | 50 | |
| IPPO2026.05 | 0 | 50 | |
| MASAC2026.05 | 0 | 50 | |
| pH-MARL2026.05 | 0 | 50 | |
| GSA2026.05 | 0 | 50 | |
| Diamond Attention (w/o mask)structured random masking=disabled2026.05 | 0 | 50 | |
| Diamond Attention (dropout)dropout=enabled2026.05 | 0 | 50 |