List your company

Question Answering on BBQ (test)

98.86Accuracy (amb)MAD
51.5463.82576.1188.395Jan 21, 2026

Evaluation results

MethodLinks
MADModel=Llama-3.1-70B-Instruct2026.0198.8681.7890.320.00550.01270.0091
MPTModel=Llama-3.1-70B-Instruct2026.0198.4687.0292.740.0030.00770.0053
Re-PromptingVariant=Debias, Model=Llama-3.1-70B-Instruct2026.0198.3380.8789.60.00630.03410.0202
Re-PromptingVariant=Persona, Model=Llama-3.1-70B-Instruct2026.0196.8671.7984.320.01040.03730.0238
Self-ConsistencyVariant=Debias, Model=Llama-3.1-70B-Instruct2026.0196.4986.2391.360.02230.01120.0168
Self-ConsistencyVariant=Standard, Model=Llama-3.1-70B-Instruct2026.0194.2992.1993.240.04240.01360.028
Direct-PromptingVariant=Debias, Model=Llama-3.1-70B-Instruct2026.0192.8279.7286.270.02110.01090.016
Direct-PromptingVariant=Standard, Model=Llama-3.1-70B-Instruct2026.0192.7388.5690.640.04060.01150.026
Self-ConsistencyVariant=Persona, Model=Llama-3.1-70B-Instruct2026.0191.2675.4383.350.01560.00630.0109
MPTModel=Llama-3.1-8B-Instruct2026.0190.4687.5589.010.02750.08490.0562
Direct-PromptingVariant=Persona, Model=Llama-3.1-70B-Instruct2026.0183.967177.480.01660.01080.0137
Re-PromptingVariant=Debias, Model=Llama-3.1-8B-Instruct2026.0183.3680.7782.070.03870.09850.0686
MADModel=Llama-3.1-8B-Instruct2026.0183.2983.1183.20.03570.11580.0758
Re-PromptingVariant=Persona, Model=Llama-3.1-8B-Instruct2026.0173.370.5571.920.03590.11640.0761
Self-ConsistencyVariant=Debias, Model=Llama-3.1-8B-Instruct2026.0164.6190.577.550.08110.07930.0802
Self-ConsistencyVariant=Standard, Model=Llama-3.1-8B-Instruct2026.0164.4891.678.040.08410.06770.0759
Direct-PromptingVariant=Standard, Model=Llama-3.1-8B-Instruct2026.0161.7589.7675.750.08390.07210.078
Direct-PromptingVariant=Debias, Model=Llama-3.1-8B-Instruct2026.0161.0388.0174.520.06760.0870.0773
Self-ConsistencyVariant=Persona, Model=Llama-3.1-8B-Instruct2026.0155.7490.1172.920.12130.07040.0958
Direct-PromptingVariant=Persona, Model=Llama-3.1-8B-Instruct2026.0153.3686.5769.960.0880.07540.0817