Goal-conditioned Reinforcement Learning on OGBench Pixel-based
95AntMaze Medium Navigate SuccessCRL
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CRLMethod Architecture=Non-Hierarchical2026.05 | 95 | 84 | 69 | 11 | 31 | 11 | 0 | 43 | |
| Ms.PRMethod Architecture=Ms.PR2026.05 | 94 | 77 | 87 | 20 | 86 | 56 | 37 | 65 | |
| HIQLMethod Architecture=Hierarchical2026.05 | 93 | 53 | 87 | 28 | 89 | 49 | 73 | 67 | |
| DualCRLMethod Architecture=Goal Representation approach (Dual)2026.05 | 85 | 43 | 20 | 1 | 14 | 15 | 0 | 25 | |
| DualGCIVLMethod Architecture=Goal Representation approach (Dual)2026.05 | 78 | 40 | 37 | 10 | 58 | 26 | 0 | 36 | |
| DualGCIQLMethod Architecture=Goal Representation approach (Dual)2026.05 | 71 | 51 | 26 | 1 | 15 | 7 | 0 | 25 | |
| GCIVLMethod Architecture=Non-Hierarchical2026.05 | 22 | 5 | 6 | 1 | 60 | 25 | 21 | 20 | |
| GCBCMethod Architecture=Non-Hierarchical2026.05 | 11 | 4 | 67 | 24 | 5 | 12 | 0 | 18 | |
| GCIQLMethod Architecture=Non-Hierarchical2026.05 | 11 | 4 | 2 | 0 | 30 | 12 | 1 | 10 | |
| QRLMethod Architecture=Non-Hierarchical2026.05 | 0 | 0 | 0 | 1 | 41 | 10 | 1 | 9 |