Learning When to Trust via Selective Context Preference Optimization

Xian Sun, Wei Chow, Yingshuo Wang +4

Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, training models to resist…