English(EN) When Confidence Signals Disagree: Local and Global Confidence in Autoregressive Language Models

语言模型的置信信号不一致，影响可靠性评估

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-16 04:00

一篇新的研究论文探讨了自回归语言模型中局部置信信号和全局置信信号之间的区别。研究发现，这两个度量——一个来自贪婪选择的答案标记的概率，另一个来自重复抽样的答案频率——之间的相关性很弱，并且与正确性的关联也不同。全局置信度与准确性之间存在中度关联，而局部置信度几乎没有相关性。研究还表明，这些置信信号之间的不一致可能预示着模型抽样不稳定，尤其是在 ARC 挑战等基准测试上。 AI

影响强调需要仔细解释模型置信度指标，影响 AI 系统的评估和控制方式。

排序理由一篇在 arXiv 上发表的研究论文，详细介绍了关于语言模型置信信号的发现。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 14 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 一篇在 arXiv 上发表的研究论文，详细介绍了关于语言模型置信信号的发现。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, model release; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Julio C. Amador Diaz Lopez · 2026-09-16 04:00

当置信信号不一致时：自回归语言模型的局部与全局置信度

arXiv:2609.16933v1 Announce Type: cross Abstract: Modern predictive systems expose multiple quantities that are commonly interpreted as measures of confidence. However, these quantities can summarize different aspects of the predictive process. This distinction matters when confi…

本文如何被排名

报道来源 [1]

当置信信号不一致时：自回归语言模型的局部与全局置信度

相关实体

相关话题