Large Multimodal Models for Low-Resource Languages: A Survey

Ana-Cristina Rogoz, Marian Lupaşcu, Mihai Sorin Stupariu, Radu Tudor Ionescu · 2025

Read Large Multimodal Models for Low-Resource Languages: A Survey by Ana-Cristina Rogoz, Marian Lupaşcu, Mihai Sorin Stupariu, Radu Tudor Ionescu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-05568.txt.

Description

Ana-Cristina Rogoz and co-authors survey how large multimodal models, which work with both image and text, are adapted for languages with little data. Such languages lack the vast text and computing stores that rich languages enjoy, so models often serve their speakers poorly.

The survey sorts 117 studies across 96 languages. It groups them by whether they add resources or change methods, and it weighs the gains and costs of each path. Images often act as a bridge that lifts results in thin data settings. The authors flag hard open problems, from made-up model claims to high computing cost.

Catalog record

CREATORAna-Cristina Rogoz, Marian Lupaşcu, Mihai Sorin Stupariu, Radu Tudor Ionescu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-05568

Browse the full library · About the Tower of Babel