XeroAlign: Zero-Shot Cross-lingual Transformer Alignment
This addresses the challenge of limited labelled data for low-resource languages in multilingual NLP, offering an incremental improvement over existing zero-shot methods.
The paper tackled the problem of zero-shot cross-lingual performance gaps in NLP by introducing XeroAlign, a method for aligning cross-lingual transformers using translated task data, resulting in state-of-the-art accuracy on three multilingual understanding tasks and competitive performance on a cross-lingual adversarial paraphrasing task.
The introduction of pretrained cross-lingual language models brought decisive improvements to multilingual NLP tasks. However, the lack of labelled task data necessitates a variety of methods aiming to close the gap to high-resource languages. Zero-shot methods in particular, often use translated task data as a training signal to bridge the performance gap between the source and target language(s). We introduce XeroAlign, a simple method for task-specific alignment of cross-lingual pretrained transformers such as XLM-R. XeroAlign uses translated task data to encourage the model to generate similar sentence embeddings for different languages. The XeroAligned XLM-R, called XLM-RA, shows strong improvements over the baseline models to achieve state-of-the-art zero-shot results on three multilingual natural language understanding tasks. XLM-RA's text classification accuracy exceeds that of XLM-R trained with labelled data and performs on par with state-of-the-art models on a cross-lingual adversarial paraphrasing task.