Renaissance Labs

Atlas / Europe

Institute · Prague, Czechia

HPLTHigh Performance Language Technologies

HPLT is an EU-funded consortium coordinated by Charles University that publishes petabyte-scale multilingual web corpora and models.

multilingual datasetsmachine translationlanguage models

Prague, Czechia

Founded
2022
Type
Institute
Based in
Prague, Czechia
Key papers
1
Featured projects
7

About

High Performance Language Technologies is a three-year EU-funded project that started in September 2022. It is coordinated by Charles University in Prague and brings together five universities, two high-performance computing centers and a private NLP company.

Its first release covered monolingual data in 75 languages at 5.6 trillion tokens and parallel data in 18 language pairs. The second release reached 7.6 trillion tokens in 193 languages and 380 million parallel sentences in 51 pairs, distributed under the CC0 license on Hugging Face and its website.

Mission and ambitions

To provide open multilingual datasets, translation models and tools for training language models at scale.

Known for

  1. 01

    Petabyte-scale multilingual web corpora

  2. 02

    HPLT v2 CC0 datasets

  3. 03

    OpusCleaner and OpusTrainer tools

Milestones

  1. 2022

    Project starts

    The EU-funded HPLT project began in September 2022 under Charles University.

  2. 2022

    OpusCleaner

    A web interface for cleaning machine translation training data was published.

  3. 2024

    First data and model release

    HPLT published version 1 datasets and models, described at EAMT 2024.

  4. 2024

    HPLT-MT models

    Configuration and scripts for HPLT translation model releases were published.

  5. 2025

    HPLT v2 data release

    The second release of cleaned monolingual and parallel data under CC0 was described at MT Summit 2025.

Open source

GitHubgithub.com/hplt-project
  • Python port of Moses tokenizer, truecaser and normalizer (498 stars)

  • OpusCleaner is a web interface that helps you select, clean and schedule your data for training machine transl (58 stars)

  • Curriculum training (25 stars)

  • HPLT Analytics (16 stars)

Hugging Facehuggingface.co/HPLT

Research papers

Notable releases

Openness

Datasets, models and tools are public (the HPLT3.0 card lists CC0).

People

Listed from the lab's own pages and public profiles.

Contact and community

Contact through the project website.

Sources

Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.

Labs nearby