Repositories
Code, datasets, and tools.
Mine
Hate and abusive-language datasets for African languages.
Machine translation of scientific text into African languages.
A multimodal Hausa dataset built from Visual Genome.
Tooling for building multimodal datasets.
Visual question answering in Hausa.
Emotion analysis with large language models for African languages.
MasaKhane's WMT22 shared-task work.
Contributed to
Corrections to the FLORES benchmark for African languages.
Summarisation of South African court judgments.