A New Massive Multilingual Dataset for High-Performance Language Technologies

TL;DR

This paper introduces HPLT, a massive multilingual dataset from web crawls, including monolingual and bilingual corpora for 75 languages, designed to advance language modeling and translation.

Contribution

The paper presents a large-scale, open multilingual dataset with novel data collection and processing methods for low-resource languages, enabling improved NLP applications.

Findings

Contains ~5.6 trillion tokens across 75 languages

Includes 96 million aligned sentence pairs for 18 language pairs

Provides one of the largest open multilingual corpora available

Abstract

We present the HPLT (High Performance Language Technologies) language resources, a new massive multilingual dataset including both monolingual and bilingual corpora extracted from CommonCrawl and previously unused web crawls from the Internet Archive. We describe our methods for data acquisition, management and processing of large corpora, which rely on open-source software tools and high-performance computing. Our monolingual collection focuses on low- to medium-resourced languages and covers 75 languages and a total of ~5.6 trillion word tokens de-duplicated on the document level. Our English-centric parallel corpus is derived from its monolingual counterpart and covers 18 language pairs and more than 96 million aligned sentence pairs with roughly 1.4 billion English tokens. The HPLT language resources are one of the largest open text corpora ever released, providing a great resource for language modeling and machine translation training. We publicly release the corpora, the software, and the tools used in this work.