List your company
ResearchFormerly WizWand

Offline Reinforcement Learning on DMControl walker-walk (expert)

97.97Normalized ScoreCQL + S2P
8.259631.549854.8478.1302Sep 30, 2022Apr 12, 2023Oct 24, 2023May 5, 2024Nov 16, 2024May 29, 2025Dec 10, 2025
Updated 8mo ago

Evaluation results

MethodLinks
CQL + S2PS2P Augmentation=true2022.0997.97
CQLS2P Augmentation=false2022.0995.43
IQL + S2PS2P Augmentation=true2022.0994.97
IQLS2P Augmentation=false2022.0994.34
LAWMaction-conditioned data percentage=5%2025.1294.3
C-LAPaction-conditioned data percentage=100%, variant=Oracle2025.1293.1
TD3+BCaction-conditioned data percentage=100%, variant=Oracle2025.1292.8
IDM-TD3+BCaction-conditioned data percentage=5%2025.1291
TD3+BCaction-conditioned data percentage=5%2025.1289.6
C-LAPaction-conditioned data percentage=5%2025.1288
SLAC-off + S2PS2P Augmentation=true2022.0970.95
SLAC-offS2P Augmentation=false2022.0911.71