Multi-Agent Reinforcement Learning on Battle N=10,000 (Mixed)
76.2Normalized ReturnMF-Diffuser
Evaluation Results
| Method | Links | |
|---|---|---|
| MF-Diffuser2026.05 | 76.2 | |
| MF-CDMs-RL2026.05 | 61.4 | |
| Oryx2026.05 | 59.7 | |
| MFQ-Offline2026.05 | 51.9 | |
| DoF2026.05 | 41.6 | |
| Indep. Diffuser2026.05 | 37.1 | |
| MA-TD3+BC2026.05 | 9.8 | |
| OMAR2026.05 | 4.9 | |
| MADiffattention=chunked-attention (max 12-agent windows)2026.05 | 2.7 | |
| Joint Diffuserattention=chunked-attention (block size 32)2026.05 | 1.4 |