arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.03528cs.CL
收藏

高效升级多语言机器翻译模型以支持更多语言

Efficiently Upgrading Multilingual Machine Translation Models to Support More Languages

University of Massachusetts Amherst(马萨诸塞大学阿默斯特分校);Meta AI

Simeng Sun, Maha Elbayad, Anna Sun, James Cross

摘要:

随着多语言机器翻译(MMT)模型的规模和支持语言的数量持续增长,随着数据以更多语言可用,重用和升级现有模型以节省计算是很自然的。然而,添加新的语言需要更新词汇表,这使得嵌入的重用变得复杂。如何在重用现有模型的同时进行架构更改以提供新旧语言的能力,这个问题也没有得到深入研究。在这项工作中,我们介绍了三种技术,帮助加快有效的学习新的语言和减轻灾难性遗忘,尽管词汇和架构不匹配。我们的研究结果表明,通过(1)仔细初始化网络,(2)应用学习率缩放,以及(3)执行数据上采样,有可能以30%的计算量超过相同大小的基线模型的性能,并恢复从头开始训练的更大模型的性能,计算量减少超过50%。此外,我们的分析表明,引入的技术有助于更有效地学习新的方向,并在同一时间减轻灾难性遗忘。我们希望我们的工作将指导研究更有效的方法来为这些MMT模型增加语言,并最终最大限度地重用现有模型。

英文摘要:

With multilingual machine translation (MMT) models continuing to grow in size and number of supported languages, it is natural to reuse and upgrade existing models to save computation as data becomes available in more languages. However, adding new languages requires updating the vocabulary, which complicates the reuse of embeddings. The question of how to reuse existing models while also making architectural changes to provide capacity for both old and new languages has also not been closely studied. In this work, we introduce three techniques that help speed up effective learning of the new languages and alleviate catastrophic forgetting despite vocabulary and architecture mismatches. Our results show that by (1) carefully initializing the network, (2) applying learning rate scaling, and (3) performing data up-sampling, it is possible to exceed the performance of a same-sized baseline model with 30% computation and recover the performance of a larger model trained from scratch with over 50% reduction in computation. Furthermore, our analysis reveals that the introduced techniques help learn the new directions more effectively and alleviate catastrophic forgetting at the same time. We hope our work will guide research into more efficient approaches to growing languages for these MMT models and ultimately maximize the reuse of existing models.

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。

↑