Multi-Agent Reinforcement Learning on Gaussian Squeeze Med-Replay N=10,000
84Normalized Return (%)MF-Diffuser
Evaluation results
| Method | Links | |
|---|---|---|
| MF-Diffuser2026.05 | 84 | |
| MF-CDMs-RL2026.05 | 76.3 | |
| Oryx2026.05 | 75 | |
| MFQ-Offline2026.05 | 67.5 | |
| DoF2026.05 | 33.8 | |
| Indep. Diffuser2026.05 | 29.5 | |
| MA-TD3+BC2026.05 | 10.5 | |
| MADiffattention=chunked-attention (max 12-agent windows)2026.05 | 6.5 | |
| OMAR2026.05 | 5.8 | |
| Joint Diffuserattention=chunked-attention (block size 32)2026.05 | 3.9 |