List your company

Safety Evaluation on SecureBreak

4.44ASRMLP
1.737619.978838.2256.4612May 6, 2026

Evaluation results

MethodLinks
MLPBase Model=Llama-3.1-70B, Condition=Cured2026.054.44
MoEBase Model=Llama-3.1-8B, Condition=Cured2026.056.25
MoEBase Model=Gemma-2-27B, Condition=Cured2026.056.62
MoEBase Model=Gemma-2-9b, Condition=Cured2026.056.94
MLPBase Model=Qwen2.5-72B, Condition=Cured2026.058.38
MoEBase Model=Mistral-7B-v0.3, Condition=Cured2026.058.56
MLPBase Model=Mistral-7B-v0.3, Condition=Cured2026.0510.44
FMBase Model=Gemma-2-9b, Condition=Cured2026.0510.62
MLPBase Model=Qwen2.5-7B, Condition=Cured2026.0510.88
MoEBase Model=Qwen2.5-7B, Condition=Cured2026.0511.31
MLPBase Model=Gemma-2-27B, Condition=Cured2026.0511.5
FMBase Model=Gemma-2-27B, Condition=Cured2026.0512.62
FMBase Model=Llama-3.1-8B, Condition=Cured2026.0517
MLPBase Model=Llama-3.1-8B, Condition=Cured2026.0518.5
MoEBase Model=Qwen2.5-72B, Condition=Cured2026.0520.31
MLPBase Model=Gemma-2-9b, Condition=Cured2026.0520.38
MoEBase Model=Llama-3.1-70B, Condition=Cured2026.0520.44
FMBase Model=Mistral-7B-v0.3, Condition=Cured2026.0525.69
FMBase Model=Qwen2.5-7B, Condition=Cured2026.0526.06
CVAEBase Model=Qwen2.5-72B, Condition=Cured2026.0531
CVAEBase Model=Gemma-2-9b, Condition=Cured2026.0531.37
FMBase Model=Qwen2.5-72B, Condition=Cured2026.0532.12
AEBase Model=Gemma-2-27B, Condition=Cured2026.0532.31
AEBase Model=Gemma-2-9b, Condition=Cured2026.0537.69
FMBase Model=Llama-3.1-70B, Condition=Cured2026.0538
AEBase Model=Qwen2.5-72B, Condition=Cured2026.0540.25
Safe LoRABase Model=Mistral-7B-v0.3, Condition=Cured2026.0540.75
CVAEBase Model=Mistral-7B-v0.3, Condition=Cured2026.0540.75
Safe LoRABase Model=Gemma-2-9b, Condition=Cured2026.0541.38
AEBase Model=Qwen2.5-7B, Condition=Cured2026.0541.81
AEBase Model=Mistral-7B-v0.3, Condition=Cured2026.0543.38
CVAEBase Model=Gemma-2-27B, Condition=Cured2026.0543.62
SaLoRABase Model=Qwen2.5-72B, Condition=Cured2026.0549.75
Safe LoRABase Model=Llama-3.1-8B, Condition=Cured2026.0549.94
Unsafe BaselineBase Model=Qwen2.5-72B, Condition=Unsafe (Finetuned)2026.0550
CVAEBase Model=Llama-3.1-70B, Condition=Cured2026.0550.25
CVAEBase Model=Llama-3.1-8B, Condition=Cured2026.0551
AEBase Model=Llama-3.1-70B, Condition=Cured2026.0551.06
AEBase Model=Llama-3.1-8B, Condition=Cured2026.0551.5
Unsafe BaselineBase Model=Gemma-2-9b, Condition=Unsafe (Finetuned)2026.0551.88
SaLoRABase Model=Gemma-2-9b, Condition=Cured2026.0551.94
Safe LoRABase Model=Qwen2.5-72B, Condition=Cured2026.0553.37
Unsafe BaselineBase Model=Gemma-2-27B, Condition=Unsafe (Finetuned)2026.0553.44
SaLoRABase Model=Mistral-7B-v0.3, Condition=Cured2026.0553.5
Unsafe BaselineBase Model=Mistral-7B-v0.3, Condition=Unsafe (Finetuned)2026.0553.87
SaLoRABase Model=Gemma-2-27B, Condition=Cured2026.0554.25
SaLoRABase Model=Llama-3.1-70B, Condition=Cured2026.0558.19
Unsafe BaselineBase Model=Qwen2.5-7B, Condition=Unsafe (Finetuned)2026.0558.81
Unsafe BaselineBase Model=Llama-3.1-70B, Condition=Unsafe (Finetuned)2026.0558.81
SaLoRABase Model=Qwen2.5-7B, Condition=Cured2026.0558.88
Safe LoRABase Model=Llama-3.1-70B, Condition=Cured2026.0559.56
Unsafe BaselineBase Model=Llama-3.1-8B, Condition=Unsafe (Finetuned)2026.0560.56
SaLoRABase Model=Llama-3.1-8B, Condition=Cured2026.0562
Safe LoRABase Model=Gemma-2-27B, Condition=Cured2026.0562.81
CVAEBase Model=Qwen2.5-7B, Condition=Cured2026.0564.44
Safe LoRABase Model=Qwen2.5-7B, Condition=Cured2026.0572