willson8972/aura-african-language-corpus

--- language: - tw license: cc-by-4.0 task_categories: - text-generation - translation - automatic-speech-recognition tags: - african-languages - twi - asante-twi - nlp - low-resource - instruction-tuning pretty_name: Aura African Language Corpus size_categories: - 1K<n<10K --- # Aura African Language Corpus Collected and curated by [Aura OS](https://auraos.uk) — an AI operating system built Africa-first. ## Dataset Description 5450 approved samples covering Asante Twi expressions, code-switching, proverbs, and instruction pairs. All community-contributed and human-reviewed. ## Languages twi_Latn ## Sample Types - **expression**: Natural Asante Twi phrases and expressions - **codeswitching**: Mixed Twi-English phrases (real conversational style) - **proverb**: Traditional Akan proverbs with English meanings - **instruction_pair**: Prompt → response pairs for fine-tuning LLMs in Asante Twi ## License CC-BY 4.0 — free to use with attribution to [Aura OS](https://auraos.uk).