okaforpascal40/BaobabAI-v0.3

🌳 BaobabAI v0.3

Africa's first continental AI model — now with 20 languages and 221,697 training pairs.

Named after the Baobab tree — the Tree of Life across Africa. Lives 2,000+ years. Feeds entire ecosystems. That is the vision.

---

🚀 What's New in v0.3

• 5 new languages added — Tigrinya, Rundi, Twi, Fulani, Malagasy, Kinyarwanda, Ewe
• 2.5x more training data — 221,697 training pairs
• 2x more training steps — 1,000 steps
• Training loss improved: 2.07 → 1.67
• New data sources: Masakhane + Glot500 + CulturaX

---

🌍 Languages Supported

Language	Country	Speakers
Hausa	Nigeria/Niger/Chad	70M+
Yoruba	Nigeria/Benin/Togo	45M+
Igbo	Nigeria	30M+
Nigerian Pidgin	West Africa	75M+
Swahili	Kenya/Tanzania	200M+
Amharic	Ethiopia	35M+
Somali	Somalia/Kenya	20M+
Xhosa	South Africa	27M+
Zulu	South Africa	27M+
Shona	Zimbabwe	15M+
Luganda	Uganda	8M+
Lingala	DRC/Congo	70M+
Wolof	Senegal/Gambia	12M+
Oromo	Ethiopia/Kenya	40M+
Tigrinya	Eritrea/Ethiopia	9M+
Rundi	Burundi/DRC	9M+
Twi	Ghana	9M+
Fulani	West Africa	40M+
Malagasy	Madagascar	25M+
Kinyarwanda	Rwanda	12M+

---

📊 Training Details

Parameter	Value
Base Model	Llama 3.2 3B Instruct
Training Pairs	221,697
Training Steps	1,000
Final Loss	1.67
Method	QLoRA (r=16) via Unsloth
Hardware	NVIDIA Tesla T4
Data Sources	Masakhane, Glot500, CulturaX

---

🔧 Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("okaforpascal40/BaobabAI-v0.3")
model = AutoModelForCausalLM.from_pretrained("okaforpascal40/BaobabAI-v0.3")

---

🗺️ Roadmap

• v0.3 ✅ — 20 languages, 221K pairs — DONE
• v0.5 🔜 — Upgrade to 8B model, 25 languages
• v1.0 🔜 — Live REST API, enterprise ready
• v2.0 🔜 — Continental dominance, 50+ languages

---

👨🏿‍💻 Built By

Pascal Okafor Ogbonna | SabiFlow Technologies Limited 🌍 baobabai.dev · GitHub

---

📜 License

Apache 2.0 — free to use, modify and build on.

---

The tree is planted. The continent will feel its shade. 🌳