XDoGE: Multilingual Data Reweighting to Enhance Language Inclusivity in LLMs

TL;DR

This paper introduces XDoGE, a multilingual data reweighting method that optimizes language data distribution to improve large language models' performance across diverse languages, especially low-resource ones.

Contribution

The paper proposes a novel multilingual reweighting algorithm, XDoGE, to better balance language data and enhance LLM performance on underrepresented languages.

Findings

Improved performance on low-resource Iberian languages.

Effective language reweighting reduces data imbalance issues.

Release of IberianLLM-7B-Instruct model trained with XDoGE.

Abstract

Current large language models (LLMs) are trained on massive amounts of text data, primarily from a few dominant languages. Studies suggest that this over-reliance on high-resource languages, such as English, hampers LLM performance in mid- and low-resource languages. To mitigate this problem, we propose to (i) optimize the language distribution by training a small proxy model within a domain-reweighing DoGE algorithm that we extend to XDoGE for a multilingual setup, and (ii) rescale the data and train a full-size model with the established language weights either from scratch or within a continual pre-training phase (CPT). We target six languages possessing a variety of geographic and intra- and inter-language-family relations, namely, English and Spanish (high-resource), Portuguese and Catalan (mid-resource), Galician and Basque (low-resource). We experiment with Salamandra-2b, which is a promising model for these languages. We investigate the effects of substantial data repetition on minor languages and under-sampling on dominant languages using the IberoBench framework for quantitative evaluation. Finally, we release a new promising IberianLLM-7B-Instruct model centering on Iberian languages and English that we pretrained from scratch and further improved using CPT with the XDoGE weights.