Aurora-M: Open Source Continual Pre-training for Multilingual Language and Code
Taishi Nakamura, Mayank Mishra, Simone Tedeschi, Yekun Chai, Jason T. Stillerman, Felix Friedrich, Prateek Yadav, Tanmay Laud
Abstract
Pretrained language models are integral part of AI applications, but their high computational cost for training limits accessibility. Initiatives such as Bloom and StarCoder aim to democratize access to pretrained models for collaborative community development. Despite these efforts, such models encounter challenges such as limited multilingual capabilities, risks of catastrophic forgetting during continual pretraining, and the high costs of training models from scratch, alongside the need to align with AI safety standards and regulatory frameworks. This paper presents Aurora-M, a 15B parameter multilingual open-source model trained on English, Finnish, Hindi, Japanese, Vietnamese, and code. Continually pretrained from StarCoderPlus on 435B additional tokens, Aurora-M surpasses 2T tokens in total training token count. It is the first open-source multilingual model fine-tuned on human-reviewed safety instructions, thus aligning its development not only with conventional red-teaming considerations, but also with the specific concerns articulated in the Biden-Harris Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. We evaluate Aurora-M across a wide range of tasks and languages, showcasing its robustness against catastrophic forgetting and its superior performance in multilingual settings, particularly in safety evaluations. We open-source Aurora-M and its variants to encourage responsible open-source development of large language models at https://huggingface.co/aurora-m.
BibTeX
@inproceedings{nakamura-etal-2025-aurora,
title = "Aurora-{M}: Open Source Continual Pre-training for Multilingual Language and Code",
author = "Nakamura, Taishi and
Mishra, Mayank and
Tedeschi, Simone and
Chai, Yekun and
Stillerman, Jason T. and
Friedrich, Felix and
Yadav, Prateek and
Laud, Tanmay and
Chien, Vu Minh and
Zhuo, Terry Yue and
Misra, Diganta and
Bogin, Ben and
Vu, Xuan-Son and
Karpinska, Marzena and
Dantuluri, Arnav Varma and
Kusa, Wojciech and
Furlanello, Tommaso and
Yokota, Rio and
Muennighoff, Niklas and
Pai, Suhas and
Adewumi, Tosin and
Laippala, Veronika and
Yao, Xiaozhe and
Junior, Adalberto Barbosa and
Drozd, Aleksandr and
Clive, Jordan and
Gupta, Kshitij and
Chen, Liangyu and
Sun, Qi and
Tsui, Ken and
Moustafa-Fahmy, Nour and
Monti, Nicolo and
Dang, Tai and
Luo, Ziyang and
Bui, Tien-Tung and
Navigli, Roberto and
Mehta, Virendra and
Blumberg, Matthew and
May, Victor and
Nguyen, Hiep and
Pyysalo, Sampo",
editor = "Rambow, Owen and
Wanner, Leo and
Apidianaki, Marianna and
Al-Khalifa, Hend and
Eugenio, Barbara Di and
Schockaert, Steven and
Darwish, Kareem and
Agarwal, Apoorv",
booktitle = "Proceedings of the 31st International Conference on Computational Linguistics: Industry Track",
month = jan,
year = "2025",
address = "Abu Dhabi, UAE",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.coling-industry.56/",
pages = "656--678"
}