Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

POLICYSHIFTBENCH

Benchmarks

Task NameDataset NameSOTA ResultTrend
Policy FollowingPOLICYSHIFTBENCH (Adaptive split)
Accuracy89
44
Policy FollowingPOLICYSHIFTBENCH Shift
Accuracy90
22
Policy-adaptive safety guardrailingPOLICYSHIFTBENCH (Shift)
F1 Score67
9
Policy-adaptive safety guardrailingPOLICYSHIFTBENCH (Adaptive split)
F1 Score86.8
9
Showing 4 of 4 rows