Developing Monolingual Setswana Datasets for Offensive Content Detection

Title: Fine-Tuning Transformer Models for Offensive Language Detection in Setswana: Code, Scripts, Explainability Assets, and Reproducibility Package Description / Release Notes This Zenodo release archives the full reproducibility package accompanying the study "Fine-Tuning Transformer Models for Semantic Detection of Cybercrime in Setswana Language." It provides all code, configuration files, and experiment artefacts required to regenerate the models, figures, tables, and explainability analyses described in the manuscript. This release adheres to the FAIR principles—ensuring that all code, documentation, and non-sensitive outputs supporting the research are findable, accessible, interoperable, and reusable. If you use this resource, please cite this work.