Offline Reinforcement Learning on D4RL
180Maze2D Large ScoreAMPL
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AMPLNumber of random seeds=52022.10 | 180 | 107.2 | 55.8 | 51.7 | 83.1 | 58.9 | 44.6 | 81.5 | 91.1 | 90.2 | 107.7 | 76 | 20.6 | 57.4 | 138.3 | 96.3 | 83.8 | 2.6 | |
| TD3+BCNumber of random seeds=52022.10 | 138.5 | 103.9 | 112.1 | 48 | 83 | 59.1 | 44.5 | 74.7 | 58.9 | 91.4 | 110.4 | 98.7 | -0.1 | -1.2 | 6.3 | 104.2 | 61.7 | 3.9 | |
| OptiDICENumber of random seeds=52022.10 | 101.7 | 119.4 | 114 | 42 | 55.7 | 57.5 | 40.5 | 37.3 | 28.1 | 66.7 | 79.4 | 52.6 | -0.9 | -0.8 | 0.4 | 86.1 | 55 | 5.5 | |
| COMBONumber of random seeds=52022.10 | 60.1 | 53.6 | 52.5 | 1.1 | 63.7 | 5.9 | 23.2 | 57.1 | -0.2 | 107.1 | 61.4 | 38.6 | 40.5 | -0.3 | 0.1 | 0.4 | 52.1 | 4.8 | |
| FisherBRCNumber of random seeds=52022.10 | 12.5 | 3.6 | -15.4 | 47.8 | 80.7 | 94.2 | 34.9 | 84.6 | 94.4 | 94.5 | 109.3 | 110 | 0.9 | 16.1 | 2.2 | 22.9 | 49.7 | 4.3 | |
| WMOPONumber of random seeds=52022.10 | 1.8 | 12.7 | -11.5 | 72 | 64.8 | 99.7 | 66.4 | 71.9 | 93.8 | 68.8 | 98.8 | 62.6 | -2.1 | -3 | 10.4 | -0.1 | 44.2 | 5 | |
| CQLNumber of random seeds=52022.10 | 1.5 | 6.3 | 37.5 | 48.6 | 81.8 | 96 | 53.8 | 80.8 | 38.7 | 64 | 110.3 | 79.3 | -1.5 | 39.5 | 119.3 | 94 | 59.6 | 4.1 | |
| MOPONumber of random seeds=52022.10 | -0.3 | 69.2 | -0.1 | 62.7 | 53.8 | -0.2 | -1.6 | 5.2 | 35.4 | 0 | 15.6 | 1.6 | 73 | 0.5 | 22.6 | 66 | 26.9 | 6.4 | |
| AlgaeDICENumber of random seeds=52022.10 | -2.2 | 2.5 | -15.3 | -0.9 | 1 | 0.9 | -3.4 | 0.5 | 1.5 | -1.8 | -0.2 | 2.2 | -3 | -2.6 | -1.3 | 0.1 | -1.4 | 8.5 |