1
Asking Is Not Enough: Protocol Sensitivity in LLM Confidence Calibration
揭露LLM置信度评估中隐含的测量偏差：概率分数与口头表达信号并非直接读取模型不确定性。
arXiv:2605.27752v1 Announce Type: new Abstract: LLM confidence calibration is often evaluated by comparing two signals: token-probability scores and v…