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

Safety Refusal Evaluation on PolyRefuse 1.0 (km)

90.9Harmful Refusal Rate

Ours (HRL + 32 LRL)

22.67640.38858.175.812May 31, 2026
Updated 1d ago

Evaluation Results

MethodLinks
2026.05
90.98.282.7
2026.05
87.287.8-0.6
2026.05
86.26.479.8
2026.05
81.1675.1
2026.05
79.95326.9
2026.05
79.2673.2
2026.05
74.315.658.7
2026.05
74.35.468.9
2026.05
741757
2026.05
70.65812.6
2026.05
63.349.613.7
2026.05
58.24.853.4
2026.05
55.813.642.2
2026.05
55.68.647
2026.05
53.123.629.5
2026.05
44.85.239.6
2026.05
43.41330.4
2026.05
40.713.227.5
2026.05
31.1724.1
2026.05
29.25.823.4
2026.05
25.33.621.7