Rollout-level failure prediction on LIBERO-OBJECT (test)
0.793AUROCSW+ATR+BO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SW+ATR+BOparameters=w*, alpha*, beta*2026.03 | 0.793 | 69.2 | |
| SW+ATRw=60, alpha=0.92026.03 | 0.772 | 76.8 | |
| Sliding Window Poolingw=702026.03 | 0.765 | 74.4 | |
| Action Transfer Reweightingalpha=0.92026.03 | 0.567 | 76.8 | |
| Baselineparams=mean_{t,d}2026.03 | 0.542 | 74.4 |