List your company

Safety and Utility Evaluation on Safety and Utility evaluation suite (test)

3HS ScoreBuffer-and-Reinforce
0.11219.60639.158.594May 23, 2026

Evaluation results

MethodLinks
Buffer-and-ReinforceBase Model Architecture=Gemma2-9B-it2026.05381.4
LDIFSBase Model Architecture=Gemma2-9B-it2026.054.181
Buffer-and-ReinforceBase Model Architecture=Average2026.055.278.3
AntidoteBase Model Architecture=Gemma2-9B-it2026.055.381
AsFTBase Model Architecture=Gemma2-9B-it2026.055.577.4
Buffer-and-ReinforceBase Model Architecture=Qwen3-4B-Instruct2026.055.580
Security VectorBase Model Architecture=Gemma2-9B-it2026.056.581.2
LisaBase Model Architecture=Gemma2-9B-it2026.057.779.7
SafeLoRABase Model Architecture=Gemma2-9B-it2026.057.878.8
Buffer-and-ReinforceBase Model Architecture=LLaMA3-8B-Instruct2026.058.477.5
AntidoteBase Model Architecture=Qwen3-4B-Instruct2026.051376.5
SafeLoRABase Model Architecture=Qwen3-4B-Instruct2026.0514.256.8
AntidoteBase Model Architecture=Average2026.0514.577.4
Security VectorBase Model Architecture=Average2026.051578.2
Security VectorBase Model Architecture=Qwen3-4B-Instruct2026.0516.981.1
LDIFSBase Model Architecture=LLaMA3-8B-Instruct2026.0517.273.8
AsFTBase Model Architecture=Average2026.0518.367.9
SafeInstructBase Model Architecture=Gemma2-9B-it2026.0518.579.9
LisaBase Model Architecture=Average2026.051976.6
SafeLoRABase Model Architecture=Average2026.0520.779.4
SafeInstructBase Model Architecture=LLaMA3-8B-Instruct2026.0521.569.8
Security VectorBase Model Architecture=LLaMA3-8B-Instruct2026.0521.772.2
SafeInstructBase Model Architecture=Average2026.0522.476.6
LisaBase Model Architecture=Qwen3-4B-Instruct2026.0522.680.9
AsFTBase Model Architecture=Qwen3-4B-Instruct2026.0523.558
AntidoteBase Model Architecture=LLaMA3-8B-Instruct2026.0525.374.7
PanaceaBase Model Architecture=Qwen3-4B-Instruct2026.0525.980.3
AsFTBase Model Architecture=LLaMA3-8B-Instruct2026.052668.3
SafeLoRABase Model Architecture=LLaMA3-8B-Instruct2026.052672.8
LisaBase Model Architecture=LLaMA3-8B-Instruct2026.0526.869.1
SafeInstructBase Model Architecture=Qwen3-4B-Instruct2026.0527.380
LDIFSBase Model Architecture=Average2026.0530.878.2
PanaceaBase Model Architecture=Gemma2-9B-it2026.053278.7
PanaceaBase Model Architecture=Average2026.053374.8
PanaceaBase Model Architecture=LLaMA3-8B-Instruct2026.054165.5
SFTBase Model Architecture=Gemma2-9B-it2026.0559.180
SFTBase Model Architecture=Average2026.0568.876
LDIFSBase Model Architecture=Qwen3-4B-Instruct2026.057179.9
SFTBase Model Architecture=Qwen3-4B-Instruct2026.0572.279.6
SFTBase Model Architecture=LLaMA3-8B-Instruct2026.0575.268.4