1 paper
Ferhat Erata, Hao Zhou, Luke Huan
We introduce fidelity probes: natural-language questions generated from a reference artifact with code-derived ground-truth answers, answered from a candidate specification. The fr…