Our Datasets and AI Tools

Browse our dataset catalog for AI training, evaluation, and benchmarking. Our datasets are available in two tiers:

Public Access: Full datasets available for download under their respective open-source licenses. Suitable for research and non-commercial use.

Commercial License: Preview samples available for review. Full datasets available for purchase for commercial AI development.

Contact services@tonative.org for pricing and licensing.

Available Datasets

Dataset Name	Description	Curation Method	Records Curated	Languages	Data Type	Validation	Access	Action	Metadata	Licensing
African Storytelling	Multilingual dataset of African contemporary and folk stories featuring 5–10 minute natural narrations for ASR, TTS, and speech-text alignment research	Commissioned	5	Igbo, Hausa, Yoruba, Dholuo	Speech + Transcript	Human QA	Commercial	Preview Sample	Croissant Metadata	CC BY-NC-ND 4.0
NaijaHealthBench	Rubric-scoring dataset for evaluating LLM performance on medical queries within the Nigerian healthcare context, and across different Nigerian languages	Adapted	500	English, Nigerian Pidgin, Igbo, Yoruba, Hausa, Fula	Parallel Text	Human QA	Commercial	Preview Sample	Croissant Metadata	CC BY-NC-ND 4.0
Swahili Parallel Text Extension	English extension of the Kidaw'ida, Kalenjin and Dholuo parallel corpus via MT and human validation	Adapted	29,230	Swahili ↔ English, Kidaw'ida, Kalenjin, Dholuo	Parallel Text - MT	AI Validated + Human QA	Public	Direct Access	—	CC-BY-NC-4.0
XNLI	Cross-lingual natural language inference for reasoning tasks	Adapted	31,500	Igbo, Kinyarwanda, Kikuyu, Luo, Yoruba, Hausa, Nigerian Pidgin	Non-Parallel Text	Human QA	Public	Direct Access	—	CC BY 4.0
MRL-Benchmark	Commonsense reasoning benchmarking dataset for LLMs	Collaborated	400	Nigerian Pidgin, Yoruba	Non-Parallel Text	Human QA	Public	Direct Access	—	CC-BY-SA-4.0

Have a question or want to get started? Reach out to our team.