@misc{indiciae20912f4bfe4f, title = {Improving Confidence Estimation in Language Models with Linear Representations of Ambiguity}, author = {Themistoklis Nikas and Nishant Subramani and Aaron Mueller}, year = {2026}, url = {https://searcharxiv.com/abs/2611.11838}, note = {Source identifier: 2611.11838} }