List your company

Bias Evaluation (Age) on BBQ Age (test)

92.5AccuracyKP
58.28467.16776.0584.933Sep 28, 2025

Evaluation results

MethodLinks
KPModel=Qwen 2.5 7B2025.0992.5—1000
KPModel=Llama 3.1 8B2025.0989.1—1000
KAPPAModel=Qwen 2.5 7B, Intervention layers=62025.0985.5—870.64
CAAModel=Qwen 2.5 7B2025.0984.3—850.67
KAPPAModel=Qwen 2.5 7B, Intervention layers=12025.0984.1—85.20.67
KAPPAModel=Qwen 2.5 7B, Intervention layers=32025.0983.9—84.90.67
Qwen 2.5 7B Instructk=32025.0983.29.3840.68
BaseModel=Qwen 2.5 7B2025.0983.2—840.68
DoLAModel=Qwen 2.5 7B2025.0982.9—83.90.68
KAPPAModel=Llama 3.1 8B, Intervention layers=62025.0976.8—81.10.31
KAPPAModel=Llama 3.1 8B, Intervention layers=12025.0973.8—750.45
KAPPAModel=Llama 3.1 8B, Intervention layers=32025.0972.3—74.40.38
CAAModel=Llama 3.1 8B2025.0965.8—64.80.53
Llama 3.1 8B Instructk=32025.0959.929.359.20.59
BaseModel=Llama 3.1 8B2025.0959.9—59.20.59
DoLAModel=Llama 3.1 8B2025.0959.6—60.10.64