1 paper · 1 filter
Gonçalo Gomes, Bruno Martins, Chrysoula Zerva
Reliable uncertainty estimation is important for deploying evaluation metrics in real-world settings, where both input ambiguity and domain shifts are common. In this work, we prop…