This repository contains the German4All dataset and code that was used to create it. The dataset and our fine-tuned model are uploaded to the HuggingFace hub for easier loading and usage. You can find further details on how to use these artifacts on their respective pages.
This repository contains the following add-ons:
- Code to reproduce the sample generation
- Code to reproduce the LLM judge annotations
- LLM judge annotations for customized filters of the synthetic dataset
- Predictions of our models on common simplification benchmark datasets (Stodden et al. 2024)
If you use any of our artifacts, please cite our paper as
@inproceedings{anschutz-etal-2025-german4all,
title = "{G}erman4{A}ll {--} A Dataset and Model for Readability-Controlled Paraphrasing in {G}erman",
author = {Ansch{\"u}tz, Miriam and
Pham, Thanh Mai and
Nasrallah, Eslam and
M{\"u}ller, Maximilian and
Craciun, Cristian-George and
Groh, Georg},
editor = "Flek, Lucie and
Narayan, Shashi and
Phương, L{\^e} Hồng and
Pei, Jiahuan",
booktitle = "Proceedings of the 18th International Natural Language Generation Conference",
month = oct,
year = "2025",
address = "Hanoi, Vietnam",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.inlg-main.24/",
pages = "390--407",
}