Statement-Tuning Enables Efficient Cross-lingual Generalization in Encoder-only Models

TL;DR

This paper demonstrates that Statement Tuning enables encoder-only models like BERT to achieve effective zero-shot cross-lingual generalization, providing a more efficient alternative to large multilingual LLMs for low-resource languages.

Contribution

The work extends Statement Tuning to multilingual NLP, showing that encoder-only models can rival multilingual LLMs in zero-shot cross-lingual tasks with improved efficiency.

Findings

Encoder models generalize well across languages

State-of-the-art encoder models rival multilingual LLMs

Efficiency gains in multilingual NLP tasks

Abstract

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.