Offline Reinforcement Learning on halfcheetah medium-replay
68.4Normalized ScoreLAWM
Evaluation Results
| Method | Links | |
|---|---|---|
| LAWMaction-conditioned data percentage=5%2025.12 | 68.4 | |
| C-LAPaction-conditioned data percentage=100%, variant=Oracle2025.12 | 65 | |
| Anti-exploration Method2026.02 | 62 | |
| Q-ALIGN DT2026.05 | 57.1 | |
| GPC-SAC2026.02 | 55.7 | |
| COMBOSeeds=52023.10 | 55.1 | |
| SAC-RND2026.02 | 54.9 | |
| QCS2026.05 | 54.1 | |
| TD3+BCaction-conditioned data percentage=100%, variant=Oracle2025.12 | 53.9 | |
| DMG2026.02 | 51.4 | |
| ProposedSeeds=52023.10 | 49.3 | |
| QT2026.05 | 48.9 | |
| TD3+BCSeeds=52023.10 | 48.3 | |
| ATACSeeds=52023.10 | 48 | |
| IDM-TD3+BCaction-conditioned data percentage=5%2025.12 | 46.6 | |
| CQLReward Specification=Step-wise2024.02 | 46.2 | |
| CQL2026.02 | 45.5 | |
| TD3-CVAE2026.02 | 45.3 | |
| TD3+BC2026.05 | 44.6 | |
| IQLSeeds=52023.10 | 44.2 | |
| IQL2026.02 | 44.2 | |
| IQL2026.05 | 44.1 | |
| O-DICE2026.02 | 44 | |
| LSDT2026.05 | 42.9 | |
| CQLSeeds=52023.10 | 41.9 | |
| DC2026.05 | 41.3 | |
| RADT2026.05 | 41.3 | |
| CGDT2026.05 | 40.4 | |
| OptiDICESeeds=52023.10 | 39.8 | |
| DM2026.05 | 39.6 | |
| BEARSeeds=52023.10 | 38.6 | |
| BCQSeeds=52023.10 | 38.2 | |
| DT2026.05 | 36.6 | |
| QDTReward Specification=Step-wise2024.02 | 35.6 | |
| C-LAPaction-conditioned data percentage=5%2025.12 | 34.6 | |
| DTReward Specification=Step-wise2024.02 | 34.1 | |
| DTReward Specification=Final Return2024.02 | 33 | |
| QDTReward Specification=Final Return2024.02 | 32.8 | |
| STCDynamics Shift=Gravity, Target Quality=medium2026.02 | 26.7 | |
| DVDF-OTDF2025.12 | 25.1 | |
| OTDFDynamics Shift=Gravity, Target Quality=medium2026.02 | 21.5 | |
| IQLDynamics Shift=Gravity, Target Quality=medium2026.02 | 20.1 | |
| BOSADynamics Shift=Gravity, Target Quality=medium2026.02 | 20 | |
| DVDF-IGDF2025.12 | 19.4 | |
| OTDF2025.12 | 17.9 | |
| DARADynamics Shift=Gravity, Target Quality=medium2026.02 | 17.6 | |
| SRPODynamics Shift=Gravity, Target Quality=medium2026.02 | 17.5 | |
| IGDF2025.12 | 14.8 | |
| IGDFDynamics Shift=Gravity, Target Quality=medium2026.02 | 14.4 | |
| TD3+BCaction-conditioned data percentage=5%2025.12 | 14 | |
| IQL2025.12 | 12.5 | |
| BOSA2025.12 | 9.5 | |
| DARA2025.12 | 8.6 | |
| CQLReward Specification=Final Return2024.02 | 7.8 |