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

Safety Controllability on Safety Domains

100Refusal Rate (Violence)

Base

0.99226.69652.478.104May 22, 2026
Updated 8d ago

Evaluation Results

MethodLinks
2026.05
100100100100100
2026.05
100100100100100
2026.05
100891.898.998.8
2026.05
1009886.712.692.8
2026.05
98.898.178.597.898.8
2026.05
98.8989.294.792.9
2026.05
95.281.197.98.596.3
2026.05
93.997.163.297.879.6
2026.05
93.993.45498.985.1
2026.05
90.395.259.196.874
2026.05
86.710090.898.916.5
2026.05
85.512.387.781.196.3
2026.05
75.994.314.368.487
2026.05
73.475.416.36181.4
2026.05
38.637.73.128.433.3
2026.05
28.994.398.950.55.6
2026.05
19.211.364.226.312.9
2026.05
15.785.898.953.618.5
2026.05
15.711.345.918.912.9
2026.05
12.110093.995.675.9
2026.05
4.89.410.22.13.7