List your company
ResearchFormerly WizWand

Predicting Language Model Output on HateBS

0.0043Brier ScoreLAMP
-0.0147560.1138720.24250.371128May 17, 2025
Updated 6mo ago

Evaluation results

MethodLinks
LAMPBase Language Model=Mistral Large2025.050.0043
LAMPBase Language Model=GPT-4.1-mini2025.050.0312
LAMPBase Language Model=Claude 3.5 Haiku2025.050.0497
LAMPBase Language Model=Gemini 2.5 Flash2025.050.072
SHAPBase Language Model=GPT-4.1-mini2025.050.0795
SHAPBase Language Model=Claude 3.5 Haiku2025.050.0815
ANCHORSBase Language Model=GPT-4.1-mini2025.050.0973
Mean baselineBase Language Model=GPT-4.1-mini2025.050.1201
LIMEBase Language Model=Claude 3.5 Haiku2025.050.1285
LIMEBase Language Model=Gemini 2.5 Flash2025.050.1489
SHAPBase Language Model=Gemini 2.5 Flash2025.050.1722
Uniform baselineBase Language Model=GPT-4.1-mini2025.050.1732
SHAPBase Language Model=Mistral Large2025.050.1732
Uniform baselineBase Language Model=Claude 3.5 Haiku2025.050.1917
LIMEBase Language Model=Mistral Large2025.050.1938
ANCHORSBase Language Model=Claude 3.5 Haiku2025.050.2043
Uniform baselineBase Language Model=Mistral Large2025.050.215
ANCHORSBase Language Model=Gemini 2.5 Flash2025.050.2189
Uniform baselineBase Language Model=Gemini 2.5 Flash2025.050.2303
ANCHORSBase Language Model=Mistral Large2025.050.2407
Random baselineBase Language Model=Claude 3.5 Haiku2025.050.2918
LIMEBase Language Model=GPT-4.1-mini2025.050.3051
Random baselineBase Language Model=GPT-4.1-mini2025.050.3409
Mean baselineBase Language Model=Claude 3.5 Haiku2025.050.3642
Random baselineBase Language Model=Mistral Large2025.050.413
Mean baselineBase Language Model=Mistral Large2025.050.4311
Random baselineBase Language Model=Gemini 2.5 Flash2025.050.4582
Mean baselineBase Language Model=Gemini 2.5 Flash2025.050.4807