Click Alarmclock Success Rate on RoboTwin Hard setting 2.0
31Success Rate (SR)OptimusVLA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OptimusVLARollouts=100, Setting=Hard2026.02 | 31 | 1 | |
| pi_0.5Rollouts=100, Setting=Hard, Reproduced=true2026.02 | 18 | 2 | |
| DP3Rollouts=100, Setting=Hard2026.02 | 14 | 3 | |
| RDTRollouts=100, Setting=Hard2026.02 | 12 | 4 | |
| pi_0Rollouts=100, Setting=Hard2026.02 | 11 | 5 | |
| DPRollouts=100, Setting=Hard2026.02 | 5 | 6 | |
| ACTRollouts=100, Setting=Hard2026.02 | 4 | 7 |