List your company

Question Answering on BBQ (Ambiguity/Disambiguation Metrics)

83.93Disambiguation TOP-1FARO-dp
69.494873.242476.9980.7376Feb 8, 2026

Evaluation results

MethodLinks
FARO-dpBackbone=Gemma-2-2b-it, Constraint=Demographic Parity (DP)2026.0283.9363.26.81-1.010.55——
Gemma-2-2b-itBackbone=Gemma-2-2b-it, Optimization=Base2026.0283.263.9114.73-0.811———
FARO-cfBackbone=Gemma-2-2b-it, Constraint=Counterfactual Fairness (CF)2026.0283.162.8610.55-0.965——0.41
FARO-eoBackbone=Gemma-2-2b-it, Constraint=Equalized Odds (EO)2026.0282.7163.7212.96-0.822—0.44—
FARO-dpBackbone=Qwen-2.5-1.5B, Constraint=Demographic Parity (DP)2026.0275.1159.189.11-0.1040.26——
Qwen-2.5-1.5BBackbone=Qwen-2.5-1.5B, Optimization=Base2026.0274.1458.9711.44-0.092———
FARO-eoBackbone=Qwen-2.5-1.5B, Constraint=Equalized Odds (EO)2026.0274.0657.6610.87-0.1—0.05—
FARO-cfBackbone=Qwen-2.5-1.5B, Constraint=Counterfactual Fairness (CF)2026.0273.1258.918.04-0.155——0.09
FARO-dpBackbone=Phi-3-Mini, Constraint=Demographic Parity (DP)2026.0271.9946.559.151.010.21——
Phi-3-MiniBackbone=Phi-3-Mini, Optimization=Base2026.0271.9242.1411.911.42———
FARO-cfBackbone=Phi-3-Mini, Constraint=Counterfactual Fairness (CF)2026.0271.7345.929.010.93——0.37
FARO-eoBackbone=Phi-3-Mini, Constraint=Equalized Odds (EO)2026.0270.0544.0110.861.04—0.18—