揭秘LLM推荐系统幻觉：模型是否自知？联合审计幻觉率与置信度校准，为目录忠实度提供新视角。
arXiv:2608.10008v1 Announce Type: cross Abstract: LLM recommenders for top-$K$ item suggestion regularly emit titles outside the target catalog. Prior…
用测试时训练校准大模型推理置信度，让 conformal 方法在分布变化下依然可靠，COLM 2026 新思路值得关注。
arXiv:2604.01170v2 Announce Type: replace-cross Abstract: While test-time scaling has enabled large language models to solve highly difficult tasks, s…
研究修剪注意力层如何影响LLM的解释忠实性与置信度校准，揭示模型优化新视角。
arXiv:2606.24970v1 Announce Type: new Abstract: Pruning Large Language Models (LLMs) reduces memory and inference costs by removing parts of the netwo…
提出潜在置信度对齐新方法，让大模型对自己的回答更有“自知之明”，值得关注。
arXiv:2606.21937v1 Announce Type: cross Abstract: Confidence calibration in large language models (LLMs) is commonly evaluated by comparing predicted …
医学影像AI的可靠性新突破：贝叶斯深度学习如何精准校准置信度与不确定性边界，值得关注。
arXiv:2602.11973v2 Announce Type: replace-cross Abstract: In critical decision support systems based on medical imaging, the reliability of AI-assiste…
利用不同模型间的分歧作为无标签正确性信号，破解自信错误检测难题。
arXiv:2603.25450v2 Announce Type: replace Abstract: Detecting when a language model is wrong without ground truth labels is a fundamental challenge fo…
ICLR 2026最新论文，提出超球面置信映射方法，高效提升深度学习模型的不确定性估计精度
arXiv:2605.05964v2 Announce Type: replace Abstract: Quantifying uncertainty in neural network predictions is essential for high-stakes domains such as…
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…
提出双重校准方法，同时校准LLM的知识与推理置信度，为提升模型可靠性提供全新路径。
arXiv:2601.11956v2 Announce Type: replace Abstract: Reliable reasoning in Large Language Models (LLMs) is challenged by their propensity for hallucina…