Edit model card

PuoBERTa-News: A Setswana Langauge Model Finetuned for News Categorisation

Zenodo doi badge arXiv 🤗 https://huggingface.co/dsfsi/PuoBERTa

Give Feedback 📑: DSFSI Resource Feedback Form{:target="_blank"}

A Roberta-based language model finetuned for News Categorisation.

Based on https://huggingface.co/dsfsi/PuoBERTa

Model Details

Model Description

This is a News Categorisation model for Setswana.

  • Developed by: Vukosi Marivate (@vukosi), Moseli Mots'Oehli (@MoseliMotsoehli) , Valencia Wagner, Richard Lastrucci and Isheanesu Dzingirai
  • Model type: RoBERTa Model
  • Language(s) (NLP): Setswana
  • License: CC BY 4.0

News Categories

We use the IPTC news codes https://iptc.org/standards/newscodes/

  1. arts_culture_entertainment_and_media (Botsweretshi, setso, boitapoloso le bobegakgang)
  2. crime_law_and_justice (Bosenyi, molao le bosiamisi)
  3. disaster_accident_and_emergency_incident (Masetlapelo, kotsi le tiragalo ya maemo a tshoganyetso)
  4. economy_business_and_finance (Ikonomi, tsa kgwebo le tsa ditšhelete)
  5. education (Thuto)
  6. environment (Tikologo)
  7. health (Boitekanelo)
  8. politics (Dipolotiki)
  9. religion_and_belief (Bodumedi le tumelo)
  10. society (Setšhaba)

Training, Dev and Validation dataset https://huggingface.co/datasets/dsfsi/daily-news-dikgang.

Model Performance

Performance of models on Daily News Dikgang dataset

Model 5-fold Cross Validation F1 Test F1
Logistic Regression + TFIDF 60.1 56.2
NCHLT TSN RoBERTa 64.7 60.3
PuoBERTa 63.8 62.9
PuoBERTaJW300 66.2 65.4


Use this model for Part of text classification for Setswana.

Citation Information

Bibtex Reference

  title   = {PuoBERTa: Training and evaluation of a curated language model for Setswana},
  author  = {Vukosi Marivate and Moseli Mots'Oehli and Valencia Wagner and Richard Lastrucci and Isheanesu Dzingirai},
  year    = {2023},
  booktitle= {Artificial Intelligence Research. SACAIR 2023. Communications in Computer and Information Science},
  url= {https://link.springer.com/chapter/10.1007/978-3-031-49002-6_17},
  keywords = {NLP},
  preprint_url = {https://arxiv.org/abs/2310.09141},
  dataset_url = {https://github.com/dsfsi/PuoBERTa},
  software_url = {https://huggingface.co/dsfsi/PuoBERTa}


Your contributions are welcome! Feel free to improve the model.

Model Card Authors

Vukosi Marivate

Model Card Contact

For more details, reach out or check our website.

Email: vukosi.marivate@cs.up.ac.za

Enjoy exploring Setswana through AI!

Downloads last month
Model size
83.5M params
Tensor type

Datasets used to train dsfsi/PuoBERTa-News