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

Agent Safety Judgment on ROME Original 1.0 (Seed)

94.53F1-Score

Abl. Unsafe

43.8356.992570.15583.3175May 5, 2026
Updated 28d ago

Evaluation Results

MethodLinks
2026.05
94.53
2026.05
93.68
2026.05
92.08
2026.05
90.59
2026.05
89.42
2026.05
88.24
2026.05
81.82
2026.05
81.48
2026.05
80.65
2026.05
80.22
2026.05
78.21
2026.05
77.51
2026.05
76.19
2026.05
74.61
2026.05
74.07
2026.05
73.27
2026.05
71.66
2026.05
66.01
2026.05
62.04
2026.05
60.14
2026.05
55.09
2026.05
51.89
2026.05
49.1
2026.05
45.78