arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.01732cs.CL
收藏

通过低资源语言进行嵌入补丁

Bootstrapping Embeddings for Low Resource Languages

School of Informatics, University of Edinburgh, UK(爱丁堡大学信息学院);Edina, University of Edinburgh, UK(爱丁堡大学埃丁纳分校)

Merve Basoz, Andrew Horne, Mattia Opper

AI总结:

本文探讨利用大语言模型生成合成三元组数据以优化嵌入模型,提出两种新方法(适配器组合和跨语言微调XL-LoRA),在多种任务和语言上取得显著提升。

摘要:

嵌入模型对现代NLP至关重要。然而,最有效的模型创建依赖于精心构建的监督微调数据。对于高资源语言如英语,此类数据 readily available。然而,对于数百种其他语言,这些数据 simply non-existent。我们研究是否大型语言模型的出现能帮助弥合这一差距。我们测试了三种不同的策略生成合成三元组数据以优化嵌入模型,包括上下文学习以及两种新方法,利用适配器组合和跨语言微调LLM生成器(XL-LoRA)分别。我们发现,尽管上下文学习仍无法达到强非合成基线,适配器组合和XL-LoRA在广泛的任务和语言上提供了显著的性能提升,为生产高性能嵌入模型提供了清晰可扩展的路径。

英文摘要:

Embedding models are crucial to modern NLP. However, the creation of the most effective models relies on carefully constructed supervised finetuning data. For high resource languages, such as English, such datasets are readily available. However, for hundreds of other languages, they are simply non-existent. We investigate whether the advent of large language models can help to bridge this gap. We test three different strategies for generating synthetic triplet data used to optimise embedding models. These include in-context learning as well as two novel approaches, leveraging adapter composition and cross lingual finetuning of the LLM generator (XL-LoRA) respectively. We find that while in-context learning still falls short of strong non-synthetic baselines, adapter composition and XL-LoRA yield strong performance gains across a wide array of tasks and languages, offering a clear, scalable pathway to producing performant embedding models for a wide variety of languages.

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。

↑