NLP
Hausa–English Translation Dataset
Parallel Hausa and English pairs for translation exercises.
Dataset description
Demonstration bitext. Pairs are short, hand-written examples for the task runner and data card. They are not a 75,000-pair commercial corpus.
- Creator
- AI HUB Data Team (demo)
- Records
- 75,000
- Format
- JSONL
- Collection period
- Demo catalogue
- Updated
- 2026-11-07
- Downloads of samples
- 0
Quality score 94/100
₦180,000 licence (demo price)
At 20% platform fee: creator ₦144,000, platform ₦36,000. This is a quote. Paystack is not connected.
Samples use a 10-minute grant. Full files are not hosted.
Quality breakdown
20% completeness + 18% accuracy + 14% consistency + 12% annotation agreement + 14% documentation + 12% duplicate health + 10% metadata.
Completeness
97%
Accuracy
94%
Consistency
93%
Duplicate rate
0.8%
Annotation agreement
91%
Documentation
90%
Metadata
88%
Data preview
Safe sample only. Restricted fields are not shown.
| id | source_hau | target_en | domain |
|---|---|---|---|
| t1 | Na gode sosai. | Thank you very much. | courtesy |
| t2 | Kudi nawa ne? | How much is it? | market |
Data card
- Purpose
- Hausa to English and English to Hausa translation.
- Collection method
- Authored demo pairs. AI-assisted flag marks that drafting help is allowed in the real workflow, not that this sample was model-generated at scale.
- Annotation method
- One reference translation per source line.
- Source
- Demo catalogue.
- Size
- 75,000 catalogue target. Sample only is hosted.
- Fields
- id, source_hau, target_en, domain
- Language
- Hausa, English
- Geography
- Nigeria
- Limitations
- No informal social-media register. No audio.
- Potential bias
- Formal written Hausa is over-represented versus spoken varieties.
- Recommended uses
- Evaluating a translation task UI and licence flow.
- Prohibited uses
- Training a production translation model and claiming benchmark results.
- Licence
- commercial
Versions
v1.0 · 2026-02-01
Demo release.
+75000 / −0 · Initial score · source_hau, target_en, domain
Report dataset
Use this for privacy, copyright, harmful content or a misleading card. Do not attach more personal data than the report needs.
You can also use the report desk.