From English-Centric to Effective Bilingual: LLMs with Custom Tokenizers for Underrepresented Languages
This work addresses fairness and quality issues for underrepresented languages in AI, though it appears incremental as it builds on existing methods for vocabulary expansion and training.
The paper tackles the problem of developing bilingual large language models for underrepresented languages by proposing a model-agnostic approach that improves language performance and reduces computational costs, as demonstrated with Ukrainian, Arabic, and Georgian.
In this paper, we propose a model-agnostic cost-effective approach to developing bilingual base large language models (LLMs) to support English and any target language. The method includes vocabulary expansion, initialization of new embeddings, model training and evaluation. We performed our experiments with three languages, each using a non-Latin script - Ukrainian, Arabic, and Georgian. Our approach demonstrates improved language performance while reducing computational costs. It mitigates the disproportionate penalization of underrepresented languages, promoting fairness and minimizing adverse phenomena such as code-switching and broken grammar. Additionally, we introduce new metrics to evaluate language quality, revealing that vocabulary size significantly impacts the quality of generated text.