通过无限采样提升大模型推理性能，动态调整计算量。
Best-of-$\infty$ -- Asymptotic Performance of Test-Time LLM Ensembling
- 用多数投票法在无限样本下优化大模型输出
- 实测显示混合多模型可超越单个模型表现
- 自适应生成策略降低计算开销，适合高精度场景
我们研究了基于多数投票的大语言模型（LLMs）最佳选择方法（best-of-N），尤其关注极限情况 N → ∞，称为 oinflower。尽管该方法在极限下表现优异，但需无穷测试时间预算。为此，我们提出一种自适应生成方案，根据答案一致性动态决定样本数 N，高效分配推理资源。此外，将框架扩展至多模型加权集成，证明此类混合模型可优于任一单个模型。最优加权策略被建模为混合整数线性规划，并可高效求解。大量实验验证了该方法的有效性。
原文摘要 · Abstract (English)
We study best-of-$N$ for large language models (LLMs) where the selection is based on majority voting. In particular, we analyze the limit $N \to \infty$, which we denote as \boinflower. While this approach achieves impressive performance in the limit, it requires an infinite test-time budget. To address this, we propose an adaptive generation scheme that selects $N$ based on answer agreement, thereby efficiently allocating inference-time computation. Beyond adaptivity, we extend the framework to weighted ensembles of multiple LLMs, showing that such mixtures can outperform any individual model. The optimal ensemble weighting is formulated and efficiently computed as a mixed-integer linear program. Extensive experiments demonstrate the effectiveness of our approach.
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。