Action Generation on Overall
32.4Rule-based ScoreOrca
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| OrcaBackbone=frozen, Action Expert=trainable from scratch2026.06 | 32.4 | 55 | 14 | 6 | 27.9 | 15.1 | 30.3 | 2.9 | |
| π0.5Backbone=frozen2026.06 | 29.4 | 54 | 14 | 5 | 26.5 | 15.3 | 26.7 | 3 | |
| V-JEPA 2.1Backbone=frozen, Action Expert=trainable from scratch2026.06 | 17 | 27 | 7 | 0 | 17.4 | 10.1 | 20.5 | 0 | |
| Qwen3.5Backbone=frozen, Action Expert=trainable from scratch2026.06 | 10.5 | 18 | 5 | 0 | 13.1 | 7.6 | 11.9 | 0 |