用置信度动态调整采样，大幅降低大模型推理成本。
Reliability-Aware Adaptive Self-Consistency for Efficient Sampling in LLM Reasoning
- 根据回答置信度决定是否继续采样，而非固定次数
- 在5个模型4个数据集上实现最佳性价比，最高省70%算力
- 适合追求高效推理的开发者和部署场景
自一致性通过多样本聚合提升推理可靠性，但带来高昂的推理开销。自适应自一致性方法虽能缓解此问题，但依赖基于计数的停止规则，对所有响应一视同仁，常导致不必要的采样。我们提出可靠性感知自适应自一致性（ReASC），将自适应采样从响应计数重构为证据充分性判断，利用响应级别的置信度实现有原则的信息聚合。ReASC分为两个阶段：单样本决策阶段，对可自信回答的实例直接判定；可靠性感知累积阶段，联合考虑响应频率与置信度进行聚合。在五个模型和四个数据集上，ReASC始终优于现有基线，实现最优的准确率-成本权衡，在3B至27B参数模型中均显著提升推理效率。以Gemma-3-4B-it在GSM8K上的表现为例，相较自一致性方法，推理成本最高降低70%且保持准确率。
原文摘要 · Abstract (English)
Self-Consistency improves reasoning reliability through multi-sample aggregation, but incurs substantial inference cost. Adaptive self-consistency methods mitigate this issue by adjusting the sampling budget; however, they rely on count-based stopping rules that treat all responses equally, often leading to unnecessary sampling. We propose Reliability-Aware Adaptive Self-Consistency (ReASC), which addresses this limitation by reframing adaptive sampling from response counting to evidence sufficiency, leveraging response-level confidence for principled information aggregation. ReASC operates in two stages: a single-sample decision stage that resolves instances confidently answerable from a single response, and a reliability-aware accumulation stage that aggregates responses by jointly leveraging their frequency and confidence. Across five models and four datasets, ReASC consistently achieves the best accuracy-cost trade-off compared to existing baselines, yielding improved inference efficiency across model scales from 3B to 27B parameters. As a concrete example, ReASC reduces inference cost by up to 70\% relative to self-consistency while preserving accuracy on GSM8K using Gemma-3-4B-it.
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。