List your company
ResearchFormerly WizWand

Offline Reinforcement Learning on D4RL Hopper Medium-Replay

100.7RewardD-QL
8.86832.70956.5580.391Feb 13, 2022Sep 20, 2022Apr 28, 2023Dec 4, 2023Jul 11, 2024Feb 16, 2025Sep 24, 2025
Updated 6mo ago

Evaluation results

MethodLinks
D-QL2024.05100.7
RTB2024.05100.4
DD2024.05100
QGPO2024.0596.9
Diffuser2023.1096.8
D2024.0596.8
RGG+planning seeds=152023.1095.8
RGGplanning seeds=152023.1095.2
CQL2023.1095
CQL2024.0595
IQL2023.1094.7
IQL2024.0594.7
MOREL2023.1093.6
IDQL2024.0592.1
TT2023.1091.5
IQLTraining timesteps=1M, Random seeds=102025.0988.46
DT2023.1082.7
Transformer2024.0579.73
DTLayers=3, Hidden Channels=256, Heads=42022.0279.7
Aaren2024.0577.87
FLOWFORMERLayers=3, Hidden Channels=256, Heads=42022.0275.5
Linear Trans.Layers=3, Hidden Channels=256, Heads=42022.0274.3
FQL+BCTraining timesteps=1M, Random seeds=102025.0973.33
MOPO2023.1067.5
ReformerLayers=3, Hidden Channels=256, Heads=42022.0266.1
PerformerLayers=3, Hidden Channels=256, Heads=42022.0264.6
cosFormerLayers=3, Hidden Channels=256, Heads=42022.0259.3
AWAC2022.0237.2
BC2022.0218.1
BC2023.1018.1
BC2024.0518.1
MBOP2023.1012.4