List your company

Comparative Reasoning on COMPS 2-Choice

8.5Δ AccuracyBatch Calibration (BC)
-0.4441.8784.26.522Feb 12, 2026

Evaluation results

MethodLinks
Batch Calibration (BC)Model=Llama3.1-8B2026.028.5——-0.173
RBCorrModel=Llama3.1-8B2026.028——-0.174
Contextual Calibration (CC)Model=Llama3.1-8B2026.026.9——-0.098
Batch Calibration (BC)Model=Llama3.1-8B-Instruct2026.020——-0.003
RBCorrModel=Llama3.1-8B-Instruct2026.020——-0.025
Contextual Calibration (CC)Model=Llama3.1-8B-Instruct2026.02-0.1——-0.016
BaselineModel=Llama3.1-8B2026.02—0.790.185—
BaselineModel=Llama3.1-8B-Instruct2026.02—0.8630.036—