List your company
ResearchFormerly WizWand

Goal-Conditioned Reinforcement Learning on manipulation-cube-single-play (test)

0.9Success RateVIB
0.07840.29170.5050.7183Oct 8, 2025Oct 18, 2025Oct 29, 2025Nov 9, 2025Nov 20, 2025Dec 1, 2025Dec 12, 2025
Updated 8mo ago

Evaluation results

MethodLinks
VIBDownstream algorithm=GCIVL, Observation type=State-based2025.100.9
Dual Goal RepresentationsDownstream algorithm=GCIVL, Observation type=State-based2025.100.89
OrigDownstream algorithm=GCIVL, Observation type=State-based2025.100.52
BYOL-γDownstream algorithm=GCIVL, Observation type=State-based2025.100.51
VIPDownstream algorithm=GCIVL, Observation type=State-based2025.100.4
TRADownstream algorithm=GCIVL, Observation type=State-based2025.100.4
CRL2025.120.32
HIQL2025.120.31
Eik-HIQL2025.120.25
Eik-HiQRL2025.120.12
QRL2025.120.11