根据置信度动态停止LLM推理，省时又准确。
CGES: Confidence-Guided Early Stopping for Efficient and Accurate Self-Consistency
- 用贝叶斯框架评估答案置信度，自动决定何时停止采样
- 在5个推理基准上平均减少58%调用次数（16.0→6.7）
- 适合追求高效推理的部署场景，尤其对慢速模型有效
大型语言模型在测试时通常需多次调用，通过多数投票聚合预测结果。尽管自一致性方法（Wang et al., 2023）有效，但需固定调用次数，且当正确答案出现频率低时表现不佳。本文提出置信度引导的早停机制（CGES），基于贝叶斯框架构建候选答案的后验分布，并在某一答案累积足够后验质量时自适应终止采样。理论证明在理想校准与现实噪声置信度条件下均具保障，且满足方向漂移假设。在五个推理基准上平均调用次数从16.0降至6.7，降幅达58%，准确率仅比自一致性低0.4个百分点。
原文摘要 · Abstract (English)
Large language models (LLMs) are often queried multiple times at test time, with predictions aggregated by majority vote. While effective, this self-consistency (Wang et al., 2023) strategy requires a fixed number of calls and fails when the correct answer is infrequent. We introduce Confidence-Guided Early Stopping (CGES), a Bayesian framework that forms posteriors over candidate answers and adaptively halts sampling once one answer accumulates enough posterior mass. We prove guarantees in both an ideal calibrated regime and a realistic noisy-confidence regime under a directional drift condition. Averaged over five reasoning benchmarks, CGES reduces the average number of calls by 58% on average (from 16.0 to 6.7) while matching its accuracy within 0.4 percentage points of self-consistency.
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。