English

Multilingual Contextualization of Large Language Models for Document-Level Machine Translation

Computation and Language 2025-08-29 v2

Abstract

Large language models (LLMs) have demonstrated strong performance in sentence-level machine translation, but scaling to document-level translation remains challenging, particularly in modeling long-range dependencies and discourse phenomena across sentences and paragraphs. In this work, we propose a method to improve LLM-based long-document translation through targeted fine-tuning on high-quality document-level data, which we curate and introduce as DocBlocks. Our approach supports multiple translation paradigms, including direct document-to-document and chunk-level translation, by integrating instructions both with and without surrounding context. This enables models to better capture cross-sentence dependencies while maintaining strong sentence-level translation performance. Experimental results show that incorporating multiple translation paradigms improves document-level translation quality and inference speed compared to prompting and agent-based methods.

Keywords

Cite

@article{arxiv.2504.12140,
  title  = {Multilingual Contextualization of Large Language Models for Document-Level Machine Translation},
  author = {Miguel Moura Ramos and Patrick Fernandes and Sweta Agrawal and André F. T. Martins},
  journal= {arXiv preprint arXiv:2504.12140},
  year   = {2025}
}

Comments

COLM 2025

R2 v1 2026-06-28T23:00:38.654Z