Multimodal Large Language Models for Low-Resource Languages: A Case Study for Basque
This work addresses the challenge of building effective MLLMs for low-resource language communities, providing a practical approach with released resources.
The paper tackled the problem of developing multimodal large language models for low-resource languages, specifically Basque, by creating training datasets and testing different LLM backbones. The results showed that only 20% Basque multimodal data yields solid performance and that a Basque-adapted backbone is not necessary.
Current Multimodal Large Language Models exhibit very strong performance for several demanding tasks. While commercial MLLMs deliver acceptable performance in low-resource languages, comparable results remain unattained within the open science community. In this paper, we aim to develop a strong MLLM for a low-resource language, namely Basque. For that purpose, we develop our own training and evaluation image-text datasets. Using two different Large Language Models as backbones, the Llama-3.1-Instruct model and a Basque-adapted variant called Latxa, we explore several data mixtures for training. We show that: i) low ratios of Basque multimodal data (around 20%) are already enough to obtain solid results on Basque benchmarks, and ii) contrary to expected, a Basque instructed backbone LLM is not required to obtain a strong MLLM in Basque. Our results pave the way to develop MLLMs for other low-resource languages by openly releasing our resources.