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

Function Deletion on HarmfulBench

-99Delta Refusal Rate (%)

CNT

-102.0532-81.4441-60.835-40.2259Mar 19, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
-9991.3495
2026.03
-98.678678.34
2026.03
-97.338373.66
2026.03
-97.337667.66
2026.03
-96.6782.3487.34
2026.03
-8787.3481.67
2026.03
-84.677970
2026.03
-81.676.877.13
2026.03
-81.3356.3449.66
2026.03
-81.271.1362.73
2026.03
-77.1358.860.34
2026.03
-73.3454.6766
2026.03
-73.3455.6766
2026.03
-72.6763.6775.67
2026.03
-7041.6735.34
2026.03
-66.3362.6761.34
2026.03
-64.335863.34
2026.03
-594530.33
2026.03
-57.6663.6765.67
2026.03
-545349.66
2026.03
-5265.6765.67
2026.03
-51.9251.4254.5
2026.03
-22.6734.3436.67