Story
arxiv_llm_reliability · Nov 16, 2026 · paper
Source brief
Stability-Compensated Threshold Calibration for Selection-Conditioned Risk Control in Large Language Models
arxiv.orgNov 16, 2026
original source linked
In brief
Large language models often produce hallucinated answers, motivating selective prediction policies that abstain on uncertain queries. However, heuristic uncertainty thresholds alone do not provide statistical control of errors among accepted predictions. Linear Expectation Constr
Continues in