List your company
ResearchFormerly WizWand

Offline Reinforcement Learning on cube-double-play OGBench 5 tasks v0

69Average Success RateValue Flows
-2.7615.8734.553.13Oct 9, 2025Oct 19, 2025Oct 29, 2025Nov 8, 2025Nov 18, 2025Nov 28, 2025Dec 8, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Value FlowsPolicy Type=Flow Policies2025.1069
CODACPolicy Type=Flow Policies2025.1061
MACModel Paradigm=Model-Based2025.1253
IQNPolicy Type=Flow Policies2025.1042
FQLModel Paradigm=Model-Free2025.1229
FQLPolicy Type=Flow Policies2025.1029
IDQLModel Paradigm=Model-Free2025.1215
FBRACPolicy Type=Flow Policies2025.1015
IFQLPolicy Type=Flow Policies2025.1014
ReBRACModel Paradigm=Model-Free2025.1212
ReBRACPolicy Type=Gaussian Policies2025.1012
IQLModel Paradigm=Model-Free2025.127
IQLPolicy Type=Gaussian Policies2025.106
FMPCModel Paradigm=Model-Based2025.123
BCPolicy Type=Gaussian Policies2025.102
C51Policy Type=Flow Policies2025.102
MOPOModel Paradigm=Model-Based2025.121
MOBILEModel Paradigm=Model-Based2025.121
LEQModel Paradigm=Model-Based2025.120