arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.13638cs.CL
收藏

利用翻译腔预训练语言模型

Pretraining Language Models Using Translationese

cse.iitb.ac.in(印度理工学院班加罗尔分校计算机科学与工程系);nict.go.jp(日本国立信息与通信技术研究所)

Meet Doshi, Raj Dabre, Pushpak Bhattacharyya

AI总结:

本文提出用机器翻译生成的翻译腔合成数据预训练低资源语言模型,通过微型模型过滤提升数据质量,在印度语言上验证其性能接近干净数据,并证明合成数据可有效缩小英语与低资源语言间的预训练差距。

摘要:

在本文中,我们探索了翻译腔作为使用机器翻译生成的合成数据在低资源语言(LRLs)语言模型(LMs)预训练中的效用。我们的简单方法包括将大量网络爬取的单语文档(干净数据)翻译成低资源语言,然后使用在少量但干净的低资源语言数据上训练的微型语言模型对翻译后的文档进行过滤。以印度语言为例,我们从头开始预训练了具有28M和85M参数的语言模型,然后针对5个下游自然语言理解(NLU)任务和4个生成式(NLG)任务进行微调。我们观察到,与在干净数据上预训练相比,在过滤后的合成数据上预训练导致NLU的相对性能下降仅为0.87%,NLG为2.35%,并且这一差距在加入少量干净数据后进一步缩小。我们还研究了合成数据过滤的影响以及合成数据生成的源语言选择。此外,在少样本设置下评估持续预训练的更大模型(如Gemma-2B和Llama-3-8B)时,我们观察到使用合成数据与使用干净数据具有竞争力。我们的研究结果表明,合成数据有望弥合英语与低资源语言之间的预训练差距。

英文摘要:

In this paper, we explore the utility of translationese as synthetic data created using machine translation for pre-training language models (LMs) for low-resource languages (LRLs). Our simple methodology consists of translating large amounts of web-crawled monolingual documents (clean) into the LRLs, followed by filtering the translated documents using tiny LMs trained on small but clean LRL data. Taking the case of Indian languages, we pre-train LMs from scratch with 28M and 85M parameters, and then fine-tune them for 5 downstream natural language understanding (NLU) and 4 generative (NLG) tasks. We observe that pre-training on filtered synthetic data leads to relative performance drops of only 0.87% for NLU and 2.35% for NLG, compared to pre-training on clean data, and this gap further diminishes upon the inclusion of a small amount of clean data. We also study the impact of synthetic data filtering and the choice of source language for synthetic data generation. Furthermore, evaluating continually pre-trained larger models like Gemma-2B and Llama-3-8B in few-shot settings, we observe that using synthetic data is competitive with using clean data. Our findings suggest that synthetic data shows promise for bridging the pre-training gap between English and LRLs.

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。

↑