Offline Reinforcement Learning on D4RL Medium-Replay
45.5HalfCheetahCQL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CQL2024.01 | 45.5 | 95 | 77.2 | 77.6 | |
| HDMI2024.01 | 44.9 | 99.6 | 80.7 | 82.6 | |
| IQL2024.01 | 44.2 | 94.7 | 73.9 | 77 | |
| DiffuserLiteBackbone=R1, Runtime per action (second)=0.0142024.01 | 42.9 | 97.8 | 84.6 | 85.1 | |
| DiffuserRuntime per action (second)=0.7492024.01 | 42.2 | 96.8 | 61.2 | 75.3 | |
| DiffuserLiteBackbone=D, Runtime per action (second)=0.022024.01 | 41.6 | 96.6 | 90.2 | 86 | |
| DiffuserLiteBackbone=R2, Runtime per action (second)=0.0082024.01 | 39.6 | 93.2 | 78.2 | 81.9 | |
| DDRuntime per action (second)=2.4352024.01 | 39.3 | 100 | 75 | 81.8 | |
| BC2024.01 | 36.6 | 18.1 | 26 | 51.9 |