← back to paper
arxiv: 2607.12447 · 2 revisions
The Computational Basis of Confidence in Large Language Models