Statement-Tuning Enables Efficient Cross-lingual Generalization in Encoder-only Models
This work addresses the problem of resource-efficient NLP for low-resource languages by providing a more inclusive and computationally cheaper alternative to memory-intensive LLMs, though it builds incrementally on existing Statement Tuning methods.
The paper tackles the challenge of enabling encoder-only models like BERT to achieve zero-shot cross-lingual generalization, a task typically dominated by large language models (LLMs), and shows that these encoders can rival multilingual LLMs in performance while being more efficient, with state-of-the-art results across languages.
Large Language Models (LLMs) excel in zero-shot and few-shot tasks, but achieving similar performance with encoder-only models like BERT and RoBERTa has been challenging due to their architecture. However, encoders offer advantages such as lower computational and memory costs. Recent work adapts them for zero-shot generalization using Statement Tuning, which reformulates tasks into finite templates. We extend this approach to multilingual NLP, exploring whether encoders can achieve zero-shot cross-lingual generalization and serve as efficient alternatives to memory-intensive LLMs for low-resource languages. Our results show that state-of-the-art encoder models generalize well across languages, rivaling multilingual LLMs while being more efficient. We also analyze multilingual Statement Tuning dataset design, efficiency gains, and language-specific generalization, contributing to more inclusive and resource-efficient NLP models. We release our code and models.