Offline Reinforcement Learning on OGBench Basic single tasks
60Scene ScoreQC
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| QCLearning Paradigm=Chunking2026.06 | 60 | 93 | 67 | 29 | 6 | 0 | 43 | |
| VALUE-FLOWSLearning Paradigm=Generative2026.06 | 60 | 95 | 61 | 21 | 48 | 3 | 48 | |
| VASTLearning Paradigm=Stitching2026.06 | 60 | 98 | 73 | 44 | 79 | 19 | 62 | |
| FQLLearning Paradigm=1-step TD2026.06 | 57 | 97 | 30 | 15 | 82 | 11 | 49 | |
| FLOQLearning Paradigm=Generative2026.06 | 57 | 96 | 49 | 31 | 93 | 29 | 59 | |
| IFQLLearning Paradigm=1-step TD2026.06 | 46 | 87 | 12 | 27 | 32 | 0 | 34 | |
| FQL-nLearning Paradigm=n-step TD2026.06 | 42 | 97 | 4 | 27 | 80 | 37 | 48 | |
| IFQL-nLearning Paradigm=n-step TD2026.06 | 32 | 82 | 4 | 12 | 27 | 0 | 26 | |
| IQLLearning Paradigm=1-step TD2026.06 | 28 | 83 | 7 | 7 | 53 | 4 | 30 |