Renaissance Labs

Atlas / Europe

Academic lab · Heidelberg, Germany

Heidelberg NLP

Heidelberg NLP is the computational linguistics group at Heidelberg University, producing benchmarks such as VALSE and graph-based language models.

NLP

Heidelberg, Germany

Type
Academic lab
Based in
Heidelberg, Germany
Key papers
2
Featured projects
4

About

The Heidelberg Natural Language Processing Group is based at the Department of Computational Linguistics, where Anette Frank is among its principal researchers. Its GitHub organization, opened in 2019, hosts code for knowledge-graph extraction, multimodal explainability and model benchmarks.

Its releases include VALSE, a benchmark for vision and language models centered on linguistic phenomena, MM-SHAP for measuring multimodal contributions, and Graph Language Models, which Moritz Plenz and Anette Frank introduced in January 2024.

Mission and ambitions

To advance language understanding and make models more faithful and explainable through open benchmarks and methods.

Known for

  1. 01

    VALSE vision-language benchmark

  2. 02

    Graph Language Models

  3. 03

    MM-SHAP and CC-SHAP explainability

Milestones

  1. 2021

    CoCo-Ex

    A tool mapping text to ConceptNet concept nodes was published.

  2. 2021

    VALSE benchmark

    Parcalabescu, Cafagna, Muradjan, Frank, Calixto and Gatt released a task-independent vision-language benchmark.

  3. 2022

    MM-SHAP

    A performance-agnostic metric for multimodal contributions was released.

  4. 2023

    CC-SHAP and ancient language models

    Code for measuring explanation faithfulness and models for ancient languages were released.

  5. 2024

    Graph Language Models

    Plenz and Frank introduced language models operating on graphs.

Open source

GitHubgithub.com/Heidelberg-NLP
  • Code for our paper "Graph Language Models" (78 stars)

  • CoCo-Ex extracts meaningful concepts from natural language texts and maps them to conjunct concept nodes in Co (59 stars)

  • Data repository for the VALSE benchmark. (40 stars)

  • Open-source repository. (36 stars)

Research papers

Notable releases

Openness

Code and benchmarks are public on GitHub.

People

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

Contact and community

Contact via site.

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 Something wrong or out of date?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