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

Jailbreak Robustness on JailbreakBench (100 behaviors, 10 categories)

23.4AE (GCG)

Qwen2.5-Instruct 0.5B

-0.3125.8441218.156Jun 9, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
23.424.318.28.528.458.79899100
2026.06
8.56011.2938.554.9100100100
2026.06
3196.636.714.914.833.2799696
2026.06
1.9233.359.129.46.915.1637484
2026.06
1.8-----6--
2026.06
1.1482.988.637.15.711.8587473
2026.06
1.1494.183.427.36.217.6587888
2026.06
1.148248.323.210.923.1679493
2026.06
0.6--51.83.29344051
2026.06
0.6-37.624.214.520.1289683