Stack red cube on green cube on single-arm SO-100
2.33Partial Success ScoreTOP-AWR
Evaluation results
| Method | Links | |
|---|---|---|
| TOP-AWRObjective=Advantage-Weighted Regression using TOPReward2026.02 | 2.33 | |
| PretrainedDescription=Base policy pretrained on 200 hours of SO-100 data2026.02 | 1.33 | |
| BCObjective=Unweighted behavior cloning baseline2026.02 | 1 |