arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.19807cs.CL
收藏

Cool-Fusion:无需训练地融合大型语言模型

Cool-Fusion: Fuse Large Language Models without Training

Sun Yat-sen University(中山大学)

Cong Liu, Xiaojun Quan, Yan Pan, Liang Lin, Weigang Wu, Xu Chen

AI总结:

本文提出无需训练的 Cool-Fusion,通过文本层面的跨模型重排序融合词表不同的异构 LLM,并在 GSM8K 等基准上验证其显著增益。

摘要:

我们关注融合两个或多个异构大型语言模型(LLM)以利用其互补优势的问题。模型融合的挑战之一是高昂的计算负载,尤其是在微调或对齐词表方面。为解决这一问题,我们提出 Cool-Fusion,这是一种简单而有效的方法,用于融合源 LLM 的知识,且不需要训练。与集成方法不同,Cool-Fusion 适用于任意一组具有不同词表的源 LLM。为克服 LLM 之间的词表差异,我们在文本层面对 LLM 进行集成,使其能够以不同粒度对彼此生成的文本进行重排序。我们在多种基准数据集上开展了大量实验。在 GSM8K 上,Cool-Fusion 将三个强大源 LLM 的准确率显著提升了 17.4%。

英文摘要:

We focus on the problem of fusing two or more heterogeneous large language models (LLMs) to leverage their complementary strengths. One of the challenges of model fusion is high computational load, specifically in fine-tuning or aligning vocabularies. To address this, we propose Cool-Fusion, a simple yet effective approach that fuses the knowledge of source LLMs, which does not require training. Unlike ensemble methods, Cool-Fusion is applicable to any set of source LLMs that have different vocabularies. To overcome the vocabulary discrepancies among LLMs, we ensemble LLMs on text level, allowing them to rerank the generated texts by each other with different granularities. Extensive experiments have been conducted across a variety of benchmark datasets. On GSM8K, Cool-Fusion increases accuracy from three strong source LLMs by a significant margin of 17.4\%.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑