Model Card for DistilBERT German Text Complexity

This model is version of distilbert-base-german-cased fine-tuned for text complexity prediction on a scale between 1 and 7.

Direct Use

To use this model, use our eval_distilbert.py script.

Training Details

The model is a fine-tuned version of the distilbert-base-german-cased and a contribution to the GermEval 2022 shared task on text complexity prediction. It was fine-tuned on the dataset by Naderi et al, 2019. For further details, visit our KONVENS paper.

Citation

Please cite our INLG 2023 paper, if you use our model. BibTeX:

@inproceedings{anschutz-groh-2022-tum,
    title = "{TUM} Social Computing at {G}erm{E}val 2022: Towards the Significance of Text Statistics and Neural Embeddings in Text Complexity Prediction",
    author = {Ansch{\"u}tz, Miriam  and
      Groh, Georg},
    booktitle = "Proceedings of the GermEval 2022 Workshop on Text Complexity Assessment of German Text",
    month = sep,
    year = "2022",
    address = "Potsdam, Germany",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.germeval-1.4",
    pages = "21--26",
}