List your company

LLM Safety Evaluation on BeaverTails

9.88Ssafe ScoreLANCE
4.35765.79137.2258.6587May 8, 2026

Evaluation results

MethodLinks
LANCEBackbone=Qwen2.5-7B2026.059.888.038.25
SelfRefineBackbone=Qwen2.5-7B2026.059.868.038.02
SysPromptBackbone=Qwen2.5-7B2026.059.7987.98
VanillaBackbone=Qwen2.5-7B2026.059.748.018.03
RepEBackbone=Qwen2.5-7B2026.059.737.988.02
InstructBackbone=Qwen2.5-7B2026.059.597.587.9
InstructBackbone=Llama3-8B2026.059.553.416.71
LANCEBackbone=Llama3.1-8B2026.059.347.798.61
SelfRefineBackbone=Llama3.1-8B2026.059.276.567.67
LANCEBackbone=Llama3-8B2026.059.237.898.69
SysPromptBackbone=Llama3.1-8B2026.059.217.567.84
InstructBackbone=Llama3.1-8B2026.059.184.177
SelfRefineBackbone=Llama3-8B2026.059.036.657.86
SysPromptBackbone=Llama3-8B2026.058.47.217.74
RepEBackbone=Llama3.1-8B2026.057.667.027.63
RepEBackbone=Llama3-8B2026.055.916.457.45
VanillaBackbone=Llama3.1-8B2026.055.886.217.28
VanillaBackbone=Llama3-8B2026.054.575.597.04