List your company

Decision Making on WikipediaToxicity

100Accuracy (vs Bayes-optimal)SFT with chain-of-thought reasoning
9597.5100102.5Mar 31, 2026

Evaluation results

MethodLinks
SFT with chain-of-thought reasoningFraming=Original, Trained=Yes2026.03100
SFT with chain-of-thought reasoningFraming=Dollar, Trained=Yes2026.03100
SFT with chain-of-thought reasoningFraming=Wording, Trained=Yes2026.03100