Bilingual Adaptation of Monolingual Foundation Models

TL;DR

This paper introduces a two-stage method for adapting monolingual LLMs to new languages, effectively balancing language retention and acquisition, demonstrated on Llama models for Arabic and Hindi.

Contribution

The study proposes a novel two-stage adaptation process involving vocabulary expansion and continual pre-training, improving cross-lingual transfer for large language models.

Findings

Significant improvement in Arabic language capabilities

Slight enhancement in English proficiency

Effective adaptation demonstrated on multiple Llama models

Abstract

We present an efficient method for adapting a monolingual Large Language Model (LLM) to another language, addressing challenges of catastrophic forgetting and tokenizer limitations. We focus this study on adapting Llama 2 to Arabic. Our two-stage approach begins with expanding the vocabulary and training only the embeddings matrix, followed by full model continual pre-training on a bilingual corpus. By continually pre-training on a mix of Arabic and English corpora, the model retains its proficiency in English while acquiring capabilities in Arabic. Our approach results in significant improvements in Arabic and slight enhancements in English, demonstrating cost-effective cross-lingual transfer. We perform ablations on embedding initialization techniques, data mix ratios, and learning rates and release a detailed training recipe. To demonstrate generalizability of this approach we also adapted Llama 3 8B to Arabic and Llama 2 13B to Hindi.