README.md

metadata

license: cc-by-nc-4.0
language:
  - sw

Inkuba-Mono: Swahili Subset

This is the Swahili subset of the Inkuba-Mono dataset, originally collected by Lelapa AI – Fundamental Research Team. It includes only the Swahili text portion extracted from Inkuba-Mono. The data is released under the CC BY-NC 4.0 license.

Dataset Details

Dataset Description

• Collected by: Lelapa AI - Fundamental Research Team
• Shared by: Lelapa AI
• Languages (NLP): Swahili (this subset only)
• License: CC BY-NC 4.0

The Inkuba-Mono corpus was originally created for training InkubaLM, a small language model focusing on multiple African languages: isiZulu, Yoruba, Swahili, isiXhosa, and Hausa. You can find details about the full Inkuba-Mono dataset in the Inkuba-Mono repository and in the associated InkubaLM paper.

Dataset Sources

The original Inkuba-Mono corpus was compiled from public repositories including:

• Hugging Face
• GitHub
• Zenodo

Swahili Statistics (from Inkuba-Mono)

• Number of sentences (in Millions): 58M
• Number of tokens: ~1B

Note: This standalone dataset contains only the Swahili portion.

Uses

You can use this Swahili subset for:

• Training or fine-tuning language models for Swahili.
• Developing Swahili NLP applications such as text classification, language modeling, or translation systems.
• Any research or educational purpose that requires monolingual Swahili data.

How to Use

from datasets import load_dataset

# Example: loading this Swahili subset from Hugging Face
data = load_dataset("Alfaxad/Inkuba-Mono-Swahili")

Citation

@article{tonja2024inkubalm,
  title={InkubaLM: A small language model for low-resource African languages},
  author={Tonja, Atnafu Lambebo and Dossou, Bonaventure FP and Ojo, Jessica and Rajab, Jenalea and Thior, Fadel and Wairagala, Eric Peter and Anuoluwapo, Aremu and Moiloa, Pelonomi and Abbott, Jade and Marivate, Vukosi and others},
  journal={arXiv preprint arXiv:2408.17024},
  year={2024}
}