List your company

Harmful Content Safety Evaluation on XsTest Harmful

0Attack Success Rate (ASR)SafePath
-0.84.61015.4Jun 15, 2026

Evaluation results

MethodLinks
SafePathBackbone Model=Qwen3-8B2026.060
ST-SBackbone Model=Qwen3-8B2026.060
ST-DBackbone Model=Qwen3-8B2026.060
BaseBackbone Model=Qwen3-32B2026.060
Star1Backbone Model=Qwen3-32B2026.060
SafePathBackbone Model=Qwen3-32B2026.060
ST-SBackbone Model=Qwen3-32B2026.060
ST-DBackbone Model=Qwen3-32B2026.060
ST-DBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.060
BaseBackbone Model=Qwen3-8B2026.060.5
Star1Backbone Model=Qwen3-8B2026.060.5
ST-SBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.063.5
ST-DBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.067.5
ST-SBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.068
SafePathBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.068.5
Star1Backbone Model=DeepSeek-R1-Distill-Llama-8B2026.069.5
BaseBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.0614
SafePathBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.0614
Star1Backbone Model=DeepSeek-R1-Distill-Llama-70B2026.0615.5
BaseBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.0620