List your company

Safety Evaluation on DirectRefusal

30.66Attack Success Rate (ASR)SafeGene
29.765235.805141.84547.8849Jun 2, 2026

Evaluation results

MethodLinks
SafeGeneBackbone=Qwen2.5-7B, Downstream Task Context=MNLI, Safety Judge=Beaver-dam-7B2026.0630.66
SafeGeneBackbone=Qwen2.5-7B, Downstream Task Context=BoolQ, Safety Judge=Beaver-dam-7B2026.0631.84
SafeGeneBackbone=Qwen2.5-7B, Downstream Task Context=AG News, Safety Judge=Beaver-dam-7B2026.0632.63
SaLoRABackbone=Qwen2.5-7B, Downstream Task Context=BoolQ, Safety Judge=Beaver-dam-7B2026.0632.76
SafeGeneBackbone=Qwen2.5-7B, Downstream Task Context=Average, Safety Judge=Beaver-dam-7B2026.0633.09
SafeMERGEBackbone=Qwen2.5-7B, Downstream Task Context=MNLI, Safety Judge=Beaver-dam-7B2026.0633.16
SaLoRABackbone=Qwen2.5-7B, Downstream Task Context=AG News, Safety Judge=Beaver-dam-7B2026.0634.21
SaLoRABackbone=Qwen2.5-7B, Downstream Task Context=Average, Safety Judge=Beaver-dam-7B2026.0634.47
SafeMERGEBackbone=Qwen2.5-7B, Downstream Task Context=BoolQ, Safety Judge=Beaver-dam-7B2026.0634.61
SafeMERGEBackbone=Qwen2.5-7B, Downstream Task Context=Average, Safety Judge=Beaver-dam-7B2026.0634.84
SafeMERGEBackbone=Qwen2.5-7B, Downstream Task Context=SST2, Safety Judge=Beaver-dam-7B2026.0635
SaLoRABackbone=Qwen2.5-7B, Downstream Task Context=SST2, Safety Judge=Beaver-dam-7B2026.0635.13
SaLoRABackbone=Qwen2.5-7B, Downstream Task Context=MNLI, Safety Judge=Beaver-dam-7B2026.0635.79
SafeMERGEBackbone=Qwen2.5-7B, Downstream Task Context=AG News, Safety Judge=Beaver-dam-7B2026.0636.58
SafeLoRABackbone=Qwen2.5-7B, Downstream Task Context=BoolQ, Safety Judge=Beaver-dam-7B2026.0636.58
SafeLoRABackbone=Qwen2.5-7B, Downstream Task Context=AG News, Safety Judge=Beaver-dam-7B2026.0637.24
SafeGeneBackbone=Qwen2.5-7B, Downstream Task Context=SST2, Safety Judge=Beaver-dam-7B2026.0637.24
Fine-tunedBackbone=Qwen2.5-7B, Downstream Task Context=MNLI, Safety Judge=Beaver-dam-7B2026.0638.03
SafeLoRABackbone=Qwen2.5-7B, Downstream Task Context=Average, Safety Judge=Beaver-dam-7B2026.0638.42
SafeLoRABackbone=Qwen2.5-7B, Downstream Task Context=SST2, Safety Judge=Beaver-dam-7B2026.0639.87
SafeLoRABackbone=Qwen2.5-7B, Downstream Task Context=MNLI, Safety Judge=Beaver-dam-7B2026.0640
Fine-tunedBackbone=Qwen2.5-7B, Downstream Task Context=BoolQ, Safety Judge=Beaver-dam-7B2026.0644.87
Fine-tunedBackbone=Qwen2.5-7B, Downstream Task Context=Average, Safety Judge=Beaver-dam-7B2026.0645.33
Fine-tunedBackbone=Qwen2.5-7B, Downstream Task Context=AG News, Safety Judge=Beaver-dam-7B2026.0645.39
Fine-tunedBackbone=Qwen2.5-7B, Downstream Task Context=SST2, Safety Judge=Beaver-dam-7B2026.0653.03