提出只在关键位置集成大模型，提升长文本生成的稳定性和速度
When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling
- 根据分词差异和概率一致性，智能选择集成位置
- 仅集成不到1%的令牌，准确率仍显著提升
- 适合追求高效高质长文本生成的研究与应用
集成大型语言模型（LLM）因其能融合各模型互补优势而受到关注。通过聚合模型的下一个词概率分布来选择词元，在短文本任务中表现良好，但在长文本生成中尚未充分探索。本文发现，若在每个词元都进行集成，反而会降低长文本生成性能。我们识别出两个决定集成位置的关键因素：不同模型间的分词差异以及其下一个词概率分布的一致性。基于此，提出SAFE（Stable And Fast LLM Ensembling）框架，通过联合考虑这两点实现选择性集成。为进一步提高稳定性，当集成分布过于平滑时，采用概率锐化策略，提升选词置信度。在MATH500和BBH等多个基准上的实验表明，SAFE在准确率和效率上均优于现有方法，即使仅对少于1%的词元进行集成也取得显著增益。
原文摘要 · Abstract (English)
Ensembling Large Language Models (LLMs) has gained attention as a promising approach to surpass the performance of individual models by leveraging their complementary strengths. In particular, aggregating models' next-token probability distributions to select the next token has been shown to be effective in various tasks. However, while successful for short-form answers, its application to long-form generation remains underexplored. In this paper, we show that using existing ensemble methods in long-form generation requires a careful choice of ensembling positions, since the standard practice of ensembling at every token often degrades performance. We identify two key factors for determining the ensembling positions: tokenization mismatch across models and consensus in their next-token probability distributions. Based on this, we propose SAFE, (Stable And Fast LLM Ensembling), a framework that selectively ensembles by jointly considering these factors. To further improve stability, we apply a probability sharpening strategy when the ensemble distribution becomes overly smooth, enabling the selection of more confident tokens during ensembling. Our experiments on diverse benchmarks, including MATH500 and BBH, demonstrate that SAFE outperforms existing methods in both accuracy and efficiency, with gains achieved even when ensembling fewer than 1% of tokens.
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。