An Empirical Study on Cross-lingual Vocabulary Adaptation for Efficient Language Model Inference

TL;DR

This study empirically evaluates five cross-lingual vocabulary adaptation methods, demonstrating significant inference speed improvements and comparable downstream performance across diverse languages and models.

Contribution

It provides the first comprehensive empirical analysis of CVA methods' impact on inference efficiency and performance in generative LLMs.

Findings

CVA methods can speed up inference by up to 271.5%.

Adapting models with balanced multilingual data maintains performance.

CVA improves inference efficiency across diverse languages and models.

Abstract

The development of state-of-the-art generative large language models (LLMs) disproportionately relies on English-centric tokenizers, vocabulary and pre-training data. Despite the fact that some LLMs have multilingual capabilities, recent studies have shown that their inference efficiency deteriorates when generating text in languages other than English. This results in increased inference time and costs. Cross-lingual vocabulary adaptation (CVA) methods have been proposed for adapting models to a target language aiming to improve downstream performance. However, the effectiveness of these methods on increasing inference efficiency of generative LLMs has yet to be explored. In this paper, we perform an empirical study of five CVA methods on four generative LLMs (including monolingual and multilingual models) across four typologically-diverse languages and four natural language understanding tasks. We find that CVA substantially contributes to LLM inference speedups of up to 271.5\%. We also show that adapting LLMs that have been pre-trained on more balanced multilingual data results in downstream performance comparable to the original models.