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

Red-teaming Safety Evaluation on StrongReject

7ASR

Meta-Llama-3.1-8B (ATTAQ-aligned)

3.305628.242853.1878.1172May 30, 2025Jul 8, 2025Aug 17, 2025Sep 25, 2025Nov 4, 2025Dec 13, 2025Jan 22, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2025.05
7111.79
2025.05
7182.14
2025.05
981.86
2025.05
10131.94
2026.01
18.27--
2025.05
19212.41
2025.05
20232.48
2026.01
20.45--
2026.01
20.77--
2025.05
23332.47
2025.05
25692.66
2026.01
34.82--
2026.01
56.87--
2026.01
57.51--
2026.01
59.11--
2026.01
60.38--
2026.01
65.18--
2026.01
66.77--
2026.01
71.57--
2026.01
73.48--
2026.01
74.44--
2026.01
75.08--
2026.01
75.4--
2026.01
77.96--
2026.01
82.43--
2026.01
87.22--
2026.01
88.5--
2026.01
90.1--
2026.01
90.42--
2026.01
91.05--
2026.01
92.01--
2026.01
99.36--