List your company

Safety Refusal Evaluation on PolyRefuse 1.0 (my)

91.3Harmful Refusal RateAdaSteer (HRL + 32 LRL)
27.02843.71460.477.086May 31, 2026

Evaluation results

MethodLinks
AdaSteer (HRL + 32 LRL)Backbone=Llama-3.1-8B-Instruct, Steering Method=AdaSteer (HRL + 32 LRL)2026.0591.392.2-0.9
Ours (HRL + 32 LRL)Backbone=Gemma-2-9B, Steering Method=HRL + 32 LRL2026.0585.15.879.3
AdaSteer (Original)Backbone=Qwen2.5-7B-Instruct, Steering Method=AdaSteer (Original)2026.058266.215.8
Ours (HRL)Backbone=Gemma-2-9B, Steering Method=HRL2026.0581.64.477.2
CAST (Original)Backbone=Llama-3.1-8B-Instruct, Steering Method=CAST (Original)2026.0577.657.420.2
CAST (Original)Backbone=Gemma-2-9B, Steering Method=CAST (Original)2026.0555.64.451.2
CAST (HRL + 32 LRL)Backbone=Gemma-2-9B, Steering Method=CAST (HRL + 32 LRL)2026.0555.44.251.2
Gemma-2-9BBackbone=Gemma-2-9B, Steering Method=None2026.0554.93.851.1
CAST (HRL + 32 LRL)Backbone=Llama-3.1-8B-Instruct, Steering Method=CAST (HRL + 32 LRL)2026.0554.526.428.1
AdaSteer (Original)Backbone=Llama-3.1-8B-Instruct, Steering Method=AdaSteer (Original)2026.0551.617.434.2
Ours (HRL + 32 LRL)Backbone=Qwen2.5-7B-Instruct, Steering Method=HRL + 32 LRL2026.055110.840.2
AdaSteer (Original)Backbone=Gemma-2-9B, Steering Method=AdaSteer (Original)2026.0550.714.636.1
AdaSteer (HRL + 32 LRL)Backbone=Gemma-2-9B, Steering Method=AdaSteer (HRL + 32 LRL)2026.0548.138.49.7
Ours (HRL + 32 LRL)Backbone=Llama-3.1-8B-Instruct, Steering Method=HRL + 32 LRL2026.0546.77.239.5
Ours (HRL)Backbone=Qwen2.5-7B-Instruct, Steering Method=HRL2026.05467.838.2
Llama-3.1-8B-InstructBackbone=Llama-3.1-8B-Instruct, Steering Method=None2026.0540.2832.2
AdaSteer (HRL + 32 LRL)Backbone=Qwen2.5-7B-Instruct, Steering Method=AdaSteer (HRL + 32 LRL)2026.0538.118.619.5
CAST (Original)Backbone=Qwen2.5-7B-Instruct, Steering Method=CAST (Original)2026.0537.110.426.7
Ours (HRL)Backbone=Llama-3.1-8B-Instruct, Steering Method=HRL2026.0533.47.625.8
CAST (HRL + 32 LRL)Backbone=Qwen2.5-7B-Instruct, Steering Method=CAST (HRL + 32 LRL)2026.0530.98.422.5
Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct, Steering Method=None2026.0529.58.421.1