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

Harmful Content Safety Evaluation on AdvBench

0.19Attack Success Rate (ASR)

Star1

Updated 1mo ago

Evaluation Results

Method	Links
Star1 2026.06		0.19
SafePath 2026.06		0.38
ST-D 2026.06		0.38
ST-S 2026.06		0.58
ST-D 2026.06		0.58
ST-S 2026.06		1.35
ST-S 2026.06		1.35
ST-D 2026.06		1.73
Base 2026.06		1.92
Base 2026.06		3.85
ST-D 2026.06		4.42
SafePath 2026.06		4.81
Star1 2026.06		6.73
ST-S 2026.06		9.62
Star1 2026.06		14.04
SafePath 2026.06		14.62
Star1 2026.06		18.65
SafePath 2026.06		20.96
Base 2026.06		26.15
Base 2026.06		35.19