Renaissance Labs

Atlas / Europe

Academic lab 路 Darmstadt, Germany

AdapterHub

AdapterHub maintains the open Adapters library, a modular, parameter-efficient fine-tuning toolkit built on Hugging Face Transformers.

parameter-efficient fine-tuningmodular learningNLP

Darmstadt, Germany

Founded
2020
Type
Academic lab
Based in
Darmstadt, Germany
Key papers
2
Featured projects
6

About

AdapterHub began in 2020 as a framework for sharing and composing adapters, small sets of added weights that stand in for full fine-tuning of a pretrained model. The original paper by Pfeiffer and colleagues was presented at EMNLP 2020, and the GitHub organization was created in April 2020.

In November 2023 the project released the unified Adapters package, replacing the earlier adapter-transformers fork, with a library paper at EMNLP 2023. The library is released under the Apache 2.0 licence and had about 2,800 GitHub stars in 2026.

Version 1.2.0, announced in May 2025, added a plugin interface for attaching adapter functionality to nearly any Transformer model. The project is run by researchers associated with TU Darmstadt and serves practitioners who need to adapt large models cheaply.

Mission and ambitions

AdapterHub aims to make parameter-efficient and modular transfer learning easy to use, share and combine. It treats adapters as reusable components that the community can publish and compose.

Known for

  1. 01

    Adapters library for Transformers

  2. 02

    Parameter-efficient fine-tuning

  3. 03

    Sharing and composing adapter modules

  4. 04

    Origin in adapter-transformers

Milestones

  1. 2020

    AdapterHub framework

    Pfeiffer et al. introduced a framework for sharing and composing adapters, presented at EMNLP 2020.

  2. 2020

    adapter-transformers

    The repository for the original adapter extension of Hugging Face Transformers was created in April 2020.

  3. 2023

    Adapters library paper

    The unified Adapters package and its library paper were released and presented at EMNLP 2023.

  4. 2025

    Adapters v1.2.0

    A plugin interface was added so adapter functionality can be attached to nearly any Transformer model.

Open source

GitHubgithub.com/adapter-hub
  • A Unified Library for Parameter-Efficient and Modular Transfer Learning (2,833 stars)

  • ARCHIVED. Please use https://docs.adapterhub.ml/huggingface_hub.html || 馃攲 A central repository collecting pre- (69 stars)

  • Research code for "What to Pre-Train on? Efficient Intermediate Task Selection", EMNLP 2021 (37 stars)

  • Research code for the paper "How Good is Your Tokenizer? On the Monolingual Performance of Multilingual Langua (27 stars)

Research papers

Notable releases

Openness

Fully open-source library and adapter checkpoints on GitHub and Hugging Face.

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