Open Language Data Initiative: Advancing Low-Resource Machine Translation for Karakalpak

TL;DR

This paper introduces new datasets and neural translation models for the low-resource Karakalpak language, demonstrating improved translation performance and supporting linguistic diversity in NLP.

Contribution

It provides the first large-scale parallel corpora and open-source neural models for Karakalpak translation, advancing low-resource language machine translation.

Findings

Neural models outperform baselines

Datasets enable better translation quality

Open-sourced models facilitate further research

Abstract

This study presents several contributions for the Karakalpak language: a FLORES+ devtest dataset translated to Karakalpak, parallel corpora for Uzbek-Karakalpak, Russian-Karakalpak and English-Karakalpak of 100,000 pairs each and open-sourced fine-tuned neural models for translation across these languages. Our experiments compare different model variants and training approaches, demonstrating improvements over existing baselines. This work, conducted as part of the Open Language Data Initiative (OLDI) shared task, aims to advance machine translation capabilities for Karakalpak and contribute to expanding linguistic diversity in NLP technologies.