List your company
ResearchFormerly WizWand

Offline Reinforcement Learning on D4RL HalfCheetah (Expert)

108.8Mean Normalized ScoreAnti-exploration Method
4.38431.49258.685.708Jun 17, 2022Feb 8, 2023Oct 3, 2023May 27, 2024Jan 19, 2025Sep 13, 2025May 8, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Anti-exploration Method2026.02108.8
EPQNumber of random seeds=42026.03107.2
EPQ2026.05107.2
SAC-RND2026.02105.8
PMDBNumber of random seeds=42026.03105.7
PMDB2026.05105.7
GPC-SAC2026.02104.7
MOBILEImplementation=Reproduced2024.11103
CQLNumber of random seeds=42026.0397.3
CQL2026.0597.3
C-LAPImplementation=Reproduced2024.1197.1
CQL2026.0296.3
DMG2026.0295.9
DMGNumber of random seeds=42026.0395.9
DMG2026.0595.9
BooT-rGeneration Strategy=Teacher-forcing2022.0695.4
TT2022.0695.3
BooT-oGeneration Strategy=Teacher-forcing2022.0695
IQL2026.0295
PLASImplementation=Reproduced2024.1194.8
BooT-rGeneration Strategy=Autoregressive2022.0694.4
MOPOImplementation=Reproduced2024.1194
PSPOAveraged across=4 random seeds2026.0593.3
BooT-oGeneration Strategy=Autoregressive2022.0692.3
RRPINumber of random seeds=42026.0390.7
AMGNumber of random seeds=42026.0389.4
ADM2026.0589.4
RAMBONumber of random seeds=42026.0379.3
RAMBO2026.0579.3
MOReLNumber of random seeds=42026.038.4
MOReL2026.058.4