List your company

Harmful Content Safety Evaluation on HexPHI

1.67Attack Success Rate (ASR)ST-D
-0.023211.405922.83534.2641Jun 15, 2026

Evaluation results

MethodLinks
ST-DBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.061.67
ST-DBackbone Model=Qwen3-8B2026.065.67
ST-SBackbone Model=Qwen3-8B2026.066.67
SafePathBackbone Model=Qwen3-8B2026.067.33
ST-SBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.067.33
ST-DBackbone Model=Qwen3-32B2026.068.33
Star1Backbone Model=Qwen3-8B2026.068.67
Star1Backbone Model=Qwen3-32B2026.069
ST-SBackbone Model=Qwen3-32B2026.0610
BaseBackbone Model=Qwen3-32B2026.0612.33
BaseBackbone Model=Qwen3-8B2026.0613.33
ST-DBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.0613.33
SafePathBackbone Model=Qwen3-32B2026.0613.33
Star1Backbone Model=DeepSeek-R1-Distill-Llama-70B2026.0619.67
ST-SBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.0620
SafePathBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.0620
Star1Backbone Model=DeepSeek-R1-Distill-Llama-8B2026.0620.67
BaseBackbone Model=DeepSeek-R1-Distill-Llama-70B2026.0623.33
SafePathBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.0624.67
BaseBackbone Model=DeepSeek-R1-Distill-Llama-8B2026.0644