发现大模型对多答案问题会系统性低估自信，提出新方法提升校准效果。
Evaluating and Calibrating LLM Confidence on Questions with Multiple Correct Answers
- 通过采样多个高概率回答，聚合语义相似的置信度。
- 多答案问题中模型准确率上升但置信度反而下降，严重失准。
- 适合关注模型可靠性与多答案任务的开发者和研究者。
置信度校准对提升大语言模型（LLM）的可靠性至关重要，但现有无需训练的方法主要针对单答案问答场景。本文发现，在存在多个有效答案时，这些方法会失效，因等效正确答案间的分歧导致置信度系统性低估。为此，我们构建了MACE基准，包含12,000个跨六个领域的事实性问题，涵盖不同数量的正确答案。在15种代表性校准方法和四个LLM家族（7B-72B参数规模）上的实验表明，随着答案数量增加，准确率上升，但估计置信度持续下降，导致混合答案数量问题出现严重失准。为解决此问题，我们提出语义置信度聚合（SCA），通过对多个高概率采样回答进行语义聚合来计算置信度。SCA在混合答案设置下达到当前最优校准性能，同时保持对单答案问题的良好校准能力。
原文摘要 · Abstract (English)
Confidence calibration is essential for making large language models (LLMs) reliable, yet existing training-free methods have been primarily studied under single-answer question answering. In this paper, we show that these methods break down in the presence of multiple valid answers, where disagreement among equally correct responses leads to systematic underestimation of confidence. To enable a systematic study of this phenomenon, we introduce MACE, a benchmark of 12,000 factual questions spanning six domains with varying numbers of correct answers. Experiments across 15 representative calibration methods and four LLM families (7B-72B) reveal that while accuracy increases with answer cardinality, estimated confidence consistently decreases, causing severe miscalibration for questions with mixed answer counts. To address this issue, we propose Semantic Confidence Aggregation (SCA), which aggregates confidence over multiple high-probability sampled responses. SCA achieves state-of-the-art calibration performance under mixed-answer settings while preserving strong calibration on single-answer questions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。