| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Policy Following | POLICYSHIFTBENCH (Adaptive split) | Accuracy89 | 44 | |
| Policy Following | POLICYSHIFTBENCH Shift | Accuracy90 | 22 | |
| Policy-adaptive safety guardrailing | POLICYSHIFTBENCH (Shift) | F1 Score67 | 9 | |
| Policy-adaptive safety guardrailing | POLICYSHIFTBENCH (Adaptive split) | F1 Score86.8 | 9 |