让大模型通过自我反馈调整信心，提升判断可靠性。
REHEARSE: Experiential Rehearsal for Verbal Confidence Calibration in Large Language Models
- 基于自身信心得分经验进行跨实例校准
- 平均ECE降低58%，准确率同步提升
- 无需训练，适合安全关键场景应用
大型语言模型常表现出与实际正确性不符的信心水平，限制了其在安全关键任务中的可靠性。现有基于提示的方法将校准视为单次推理问题，依赖实例级推理或事后自评估。本文提出Rehearse（经验重演），一种无需训练的方法，使模型能从自身带有评分的信心经历中学习。在基于严格合适评分规则的可信度校准游戏中，模型会获得先前信心决策的反馈；该经验被总结为赛后轨迹前缀，捕捉系统性高估或低估行为。推理时，模型将此跨实例校准信号应用于每个新问题的思维链推理过程。在四个大模型、三个基准和五个随机种子下，Rehearse在无训练方法中实现最低平均ECE，相比未校准基线平均降低58%，且准确率提高。代码已公开于https://anonymous.4open.science/r/Experiential-Rehearsal-7C77/。
原文摘要 · Abstract (English)
Large language models (LLMs) often express verbal confidence that is poorly aligned with actual correctness, limiting their reliability in safety-critical applications. Existing prompt-based methods treat calibration largely as a one-shot inference problem, relying on either instance-level reasoning or post-hoc self-assessment. We introduce Rehearse (Experiential Rehearsal), a training-free method that instead enables models to adapt from their own scored confidence experience. In a credence-calibration game grounded in a strictly proper scoring rule, the model receives feedback on prior confidence decisions; this experience is summarized in a post-game trajectory prefix that captures systematic over- or under-confidence. At inference time, the model applies this cross-instance calibration signal to the chain-of-thought reasoning trace for each new question. Across four LLMs, three benchmarks, and five random seeds, Rehearse achieves the lowest average ECE among training-free methods with improved accuracy, reducing average ECE by 58% relative to the uncalibrated baseline. Code is available at https://anonymous.4open.science/r/Experiential-Rehearsal-7C77/.
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。