Can supervised and self-ensembled in-context learning improve task performance and model calibration?

Original title: On Task Performance and Model Calibration with Supervised and Self-Ensembled In-Context Learning Authors: Chengzu Li, Han Zhou, Goran Glavaš, Anna Korhonen, Ivan Vulić In this article, the authors explore the issue of overconfidence…

Read more of Can supervised and self-ensembled in-context learning improve task performance and model calibration?