ResearchBenchmarksOffline-to-online Reinforcement Learning on OGBench cube-triple 5 tasksFollow90Online Success RateAQC-3.620.74569.3May 7, 2026Evaluation ResultsMethodMethodLinksOnline Success RateOffline Success RateAQCTD Type=OursTD Type=Ours2026.059013QCTD Type=Q-chunkingTD Type=Q-chunking2026.05645QC-FQLTD Type=Q-chunkingTD Type=Q-chunking2026.05533RLPDTD Type=1-step TDTD Type=1-step TD2026.0541—FQLTD Type=1-step TDTD Type=1-step TD2026.05182FQL-nTD Type=n-step TDTD Type=n-step TD2026.0511IQLTD Type=1-step TDTD Type=1-step TD2026.0500DQCTD Type=Q-chunkingTD Type=Q-chunking2026.0501