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

Safety Alignment on XSTest

95.2Compliance

Yi-VL-6B

21.15240.37659.678.824Apr 14, 2025Jun 14, 2025Aug 15, 2025Oct 16, 2025Dec 16, 2025Feb 16, 2026Apr 19, 2026
Updated 22d ago

Evaluation Results

MethodLinks
2025.04
95.241.5--
2025.04
94.858.5--
2025.04
9275.5--
2025.04
91.265.5--
2025.04
90.482--
2025.04
90.490.5--
2025.04
9084.5--
2025.04
89.267--
2025.04
87.689.5--
2025.04
84.493--
2026.04
80.5-16.8-
2025.04
77.697--
2026.04
77.5-17.2-
2025.04
77.296.5--
2026.04
76.5-10-
2026.04
74-9.6-
2026.04
68.5-10.8-
2026.04
68-6-
2026.04
4456--
2026.04
2575--
2026.04
2476--
2026.05
---71.93
2026.05
---72.3
2026.05
---74.96
2026.05
---74.74
2026.05
---66.22
2026.05
---67.18
2026.05
---67.78
2026.05
---73.78
2026.05
---74
2026.05
---79.78
2026.05
---68.82
2026.05
---69.63
2026.05
---67.6
2026.05
---70.59
2026.05
---69.63