---
license: apache-2.0
datasets:
  - Ijwi-ry-Ikirundi-AI/Kirundi_Open_Speech_Dataset
language:
  - rn
  - fr
  - en
metrics:
  - bleu
base_model:
  - facebook/nllb-200-distilled-600M
pipeline_tag: translation
library_name: transformers
tags:
  - nllb
  - fine-tuned
  - kirundi
  - multilingual
  - umuseke-w-ijwi
  - ijwi-ry-ikirundi
  - burundi
model-index:
  - name: nllb-kirundi-multi
    results:
      - task:
          type: translation
        dataset:
          name: Kirundi Open Speech Dataset
          type: Ijwi-ry-Ikirundi-AI/Kirundi_Open_Speech_Dataset
        metrics:
          - name: BLEU
            type: bleu
            value: 9.94
---

<div align="center">
  <p align="center">
    <img src="https://raw.githubusercontent.com/Sama-ndari/kirundi-contribution-app/main/static/icon.png" alt="Ijwi ry'Ikirundi AI Logo" width="150" style="border-radius: 50%;">
  </p>
  <h1>nllb-kirundi-multi (Umuseke w'Ijwi) 🇧🇮</h1>
  
  <p>
    <b>Ikirundi cacu, Ijwi ryacu!</b><br>
    <i>(Our language, our voice!)</i>
  </p>

  <p>
    Fine-tuned <b>NLLB-200-600M</b> model for multidirectional translation.<br>
    Part of the <b>Umuseke w'Ijwi</b> initiative for African Linguistic Sovereignty.
  </p>

  <a href="https://samandari.dev/kirundi-contribution-app/">
    <img src="https://img.shields.io/badge/🚀_Contribute_Now-Web_App-green?style=for-the-badge&logo=html5" alt="Contribute">
  </a>
  
  <a href="https://github.com/Ijwi-ry-Ikirundi-AI">
    <img src="https://img.shields.io/badge/Source_Code-GitHub-black?style=for-the-badge&logo=github" alt="GitHub">
  </a>
</div>

---

## 🌍 Vision
Our mission is to bridge the **"Digital Linguistic Divide"** for the 30 million Kirundi speakers in East Africa. By creating open-source AI infrastructure, we ensure that Kirundi is a first-class citizen in the global digital revolution. 

This model serves as a foundational "engine" for:
- 🎙️ **Speech-to-Text** (ASR)
- 🗣️ **Text-to-Speech** (TTS)
- 🌐 **Machine Translation** (MT)

## 🚀 Key Features
This is a **multidirectional translation model** (Seq2Seq) capable of translating:
* **Kirundi (run_Latn) ↔ French (fra_Latn)**
* **Kirundi (run_Latn) ↔ English (eng_Latn)**

## 📊 Training & Performance
* **Base Model:** Meta NLLB-200-distilled-600M.
* **Dataset:** [Kirundi Open Speech Dataset](https://huggingface.co/datasets/Ijwi-ry-Ikirundi-AI/Kirundi_Open_Speech_Dataset) (approx. 11k pairs in this version).
* **Current Score:** **9.94 BLEU** (Iteration V1). Performance is expected to increase as the community adds more validated "Gold" phrases.

## 🛠️ Usage (Python)
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Ijwi-ry-Ikirundi-AI/nllb-kirundi-multi")
model = AutoModelForSeq2SeqLM.from_pretrained("Ijwi-ry-Ikirundi-AI/nllb-kirundi-multi")

# Example: Kirundi -> French
tokenizer.src_lang = "run_Latn"
text = "Amahoro y'Imana abane nawe."

inputs = tokenizer(text, return_tensors="pt")
output = model.generate(
    **inputs, 
    forced_bos_token_id=tokenizer.convert_tokens_to_ids("fra_Latn")
)

print(tokenizer.decode(output[0], skip_special_tokens=True))
```

## 🫱🏽‍🫲🏻 Community & Open Source
This project is dedicated to the **Open Source** community. We believe that linguistic data and AI models should be public goods to foster innovation and inclusion across Africa.

---

### 👥 Contributors
* **Samandari**: [github.com/Sama-ndari](https://github.com/Sama-ndari)
* **Rugwe**: [github.com/Irumva17](https://github.com/Irumva17)

*Built by [Ijwi ry'Ikirundi AI](https://huggingface.co/Ijwi-ry-Ikirundi-AI).*