Multi-Agent Reinforcement Learning on Ising Model N=10,000 (Expert)
95.8Normalized ReturnMFQ-Offline
Evaluation Results
| Method | Links | |
|---|---|---|
| MFQ-Offline2026.05 | 95.8 | |
| MF-Diffuser2026.05 | 95.3 | |
| Oryx2026.05 | 93.4 | |
| MF-CDMs-RL2026.05 | 92.8 | |
| DoF2026.05 | 84.6 | |
| Indep. Diffuser2026.05 | 79.8 | |
| MA-TD3+BC2026.05 | 32.4 | |
| OMAR2026.05 | 28.5 | |
| MADiffattention=chunked-attention (max 12-agent windows)2026.05 | 18.4 | |
| Joint Diffuserattention=chunked-attention (block size 32)2026.05 | 12.3 |