Global-Data-Science-Institute/project-tafsiri

Bridging African Languages and Modern AI Technology # Project Tafsiri **Bridging African Languages and Modern AI Technology** Project Tafsiri (Swahili for "translation") is an open-source initiative to create comprehensive, culturally-aware AI systems for African languages. We're building translation models, chatbots, and language tools that understand cultural context, preserve traditional knowledge, and empower communities to access modern technology without abandoning their linguistic heritage. ## Vision The vision of Project Tafsiri extends beyond simple translation. We are creating AI systems that: - **Understand cultural context** and linguistic nuances - **Preserve traditional knowledge** embedded in oral traditions - **Empower communities** to engage with technology in their native languages - **Maintain cultural authenticity** through community involvement - **Support linguistic diversity** across the African continent ## Current Status: LuhyaAI We're starting with **Luhya**, a Bantu language cluster spoken by over 6 million people across Kenya, Uganda, and Tanzania. Luhya provides an ideal foundation with: - **Significant speaker population** (6+ million native speakers) - **Rich dialect diversity** (Bukusu, Maragoli, Wanga, Tsotso, Kisa, and others) - **Cultural richness** with extensive oral traditions and proverbs - **Community engagement** ensuring linguistic and cultural accuracy ### Supported Luhya Dialects - Bukusu (luy_bukusu) - Wanga (luy_wanga) - Kisa (luy_kisa) - Maragoli (luy_maragoli) - Tachoni (luy_tachoni) - Kabras (luy_kabras) ## Resources ### HuggingFace Hub - **Dataset**: luhya-multilingual-dataset - 26K+ translation pairs - **Model**: luhya-multilingual-m2m100 - Fine-tuned M2M100 model - **Checkpoints**: Private repository for training checkpoints ### Research & Development - Translation model based on Facebook's M2M100 - Cultural context integration - Dialect-specific adaptations - Community validation framework ## Quick Start ### Installation ```bash git clone githu …