Robot Policy Mode Editing (4 to 2) on Push-Pillars
72.3SR (Avg)Filtered-retrain ref.
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Filtered-retrain ref.Backbone=π0.5 VLA, Mode Change=4 → 2, Evaluation Seeds=502026.06 | 72.3 | 88 | 74.7 | 88 | |
| MoREBackbone=π0.5 VLA, Mode Change=4 → 2, Evaluation Seeds=502026.06 | 70.3 | 76 | 88 | 94 | |
| Original policyBackbone=π0.5 VLA, Mode Change=4 → 2, Evaluation Seeds=502026.06 | 41 | 48 | 82 | 82 |