Key ingredients for effective zero-shot cross-lingual knowledge transfer in generative tasks: learning rate is (almost) all you need

Anonymous

TL;DR: We show that learning rate tuning is a defining factor for effective zero-shot cross-lingual generation and find other ingredients to achieve the performance of data translation

Abstract: Zero-shot cross-lingual generation implies finetuning of the multilingual pretrained language model (mPLM) on a generation task in one language and then using it to make predictions for this task in other languages. Previous works notice a frequent problem of generation in a wrong language and propose approaches to address it, usually using mT5 as a backbone model. In this work we compare various approaches proposed in the literature in unified settings, also including alternative backbone models, namely mBART and NLLB-200. We first underline the importance of tuning learning rate used for finetuning, which helps to substantially alleviate the problem of generation in the wrong language. Then, we show that with careful learning rate tuning, the simple full finetuning of the model acts as a very strong baseline and alternative approaches bring only marginal improvements. Finally, we find that mBART performs similarly to mT5 of the same size, and NLLB-200 can be competitive in some cases. Our final models reach the performance of the approach based on data translation which is usually considered as an upper baseline for zero-shot cross-lingual generation.

Paper Type: long

Research Area: Multilinguality and Language Diversity

Contribution Types: NLP engineering experiment

Languages Studied: Arabic, Chinese, German, French, English, Korean, Russian, Spanish