List your company
ResearchFormerly WizWand

Offline Reinforcement Learning on kitchen mixed

77.8Normalized ScoreWGCSL
-0.51219.81940.1560.481Jan 30, 2023Apr 2, 2023Jun 3, 2023Aug 4, 2023Oct 5, 2023Dec 6, 2023Feb 7, 2024
Updated 8mo ago

Evaluation results

MethodLinks
WGCSL2023.0777.8
GCPC2023.0775.6
GCIQL2023.0774.6
DWSL2023.0774.6
RvS-G2023.0769.4
DD2023.0765
SPQRBase algorithm=CQL-Min2024.0155.6
IQL2023.0753.2
DPPOFramework=Preference Transformer (PT)2023.0152.5
DPPOsupervision=preference only, seeds=52023.0152.5
CQL2023.0752.4
CQLReward Specification=Step-wise2024.0251
IQL2024.0151
CQL-Min2024.0151
DT2023.0750.7
IQLsupervision=task rewards, seeds=52023.0150.6
BC2023.0748.9
PT+IQLFramework=Preference Transformer (PT)2023.0148
PT+IQLsupervision=preference only, seeds=52023.0148
BC2024.0147.5
BEAR2024.0147.2
PT+%BCFramework=Preference Transformer (PT)2023.0140.9
PT+RvSFramework=Preference Transformer (PT)2023.0127.5
DTReward Specification=Step-wise2024.0222.3
DTReward Specification=Final Return2024.0217.2
PT+CQLFramework=Preference Transformer (PT)2023.0112.3
PT+CQLsupervision=preference only, seeds=52023.0112.3
CQLsupervision=task rewards, seeds=52023.0110.7
SAC-Min2024.012.5