Offline Reinforcement Learning on Ising Model N=1000 (Expert)
94.3Normalized ReturnMFQ-Offline
Evaluation results
| Method | Links | |
|---|---|---|
| MFQ-OfflineN=1000, Seeds=5, Online rollouts=102026.05 | 94.3 | |
| MF-DiffuserN=1000, Seeds=5, Online rollouts=102026.05 | 93.5 | |
| OryxN=1000, Seeds=5, Online rollouts=102026.05 | 92.5 | |
| MF-CDMs-RLN=1000, Seeds=5, Online rollouts=102026.05 | 91.2 | |
| DoFN=1000, Seeds=5, Online rollouts=102026.05 | 86.3 | |
| Indep. DiffuserN=1000, Seeds=5, Online rollouts=102026.05 | 82.6 | |
| MADiffN=1000, Seeds=5, Online rollouts=102026.05 | 76.5 | |
| Joint DiffuserN=1000, Seeds=5, Online rollouts=102026.05 | 71.3 | |
| OMARN=1000, Seeds=5, Online rollouts=10, mean-field adaptation=true2026.05 | 68.4 | |
| MA-TD3+BCN=1000, Seeds=5, Online rollouts=10, mean-field adaptation=true2026.05 | 65.2 |