Metraipolis › archive › dc_b009edbf06d5
How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates — arXiv:2610.40516
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.40516
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:14:57.945Z
- SHA-256 of the archived text
e7980e9c551ac6e61d94df6be6b7a86b970f73fee0711eca79338e951c0fb57b- Stable URL of this page
https://metraipolis.com/documento/dc_b009edbf06d5
How to cite
How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates — arXiv:2610.40516 — arXiv (en), https://arxiv.org/abs/2610.40516 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b009edbf06d5 (Metraipolis archive dc_b009edbf06d5, sha256 e7980e9c551ac6e61d94df6be6b7a86b970f73fee0711eca79338e951c0fb57b)
BibTeX
@misc{dcb009edbf06d5,
title = {How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates — arXiv:2610.40516},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.40516},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b009edbf06d5; sha256 e7980e9c551ac6e61d94df6be6b7a86b970f73fee0711eca79338e951c0fb57b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40516. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Safety alignment for Large Language Models is heavily focused on English with limited multilingual representation, particularly damaging low-resource language performance. Related open training datasets are often limited in size or do not follow region-localized harmful topics to broaden the scope of alignment beyond cultures. We propose a scalable template-based augmentation pipeline to produce massive multilingual refusal and benign corpora, extracting safe templates from harmful prompts, translating them and then filling in the slots with culturally authentic entities, both harmful and benign. As a result, we release MSafeTemp, a validated corpus spanning 60 languages useful for multilingual safety alignment and evaluation. Using the dataset, we show that English-only evaluation misjudges multilingual safety: for several models, attacks that fail in English succeed once translated, while culturally localized variants expose further failures that translation alone misses. We also confirm that augmenting safety training data through templates allows for higher diversity and quality of alignment, resulting in improved safety performance beyond English. MSafeTemp and the respective pipeline are aimed at facilitating research and development of safer and more responsible models for less-represented cultures. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.40516 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T06:07:08.777Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).