Integrating the Wolof translation model from GalsenAI
# French-Wolof Translator
A modular, production-ready French-Wolof translation system built on Facebook's NLLB (No Language Left Behind) model. This project provides a complete framework for training, evaluating, and using bidirectional translation models between French and Wolof.
## 📋 Table of Contents
- Features
- Installation
- Quick Start
- Project Structure
- Usage
- Training
- Configuration
- Versioning
- Contributing
- License
> 📖 **For detailed configuration instructions, see CONFIGURATION.md**
## ✨ Features
- **Bidirectional Translation**: Translate between French and Wolof in both directions
- **Modular Architecture**: Clean, maintainable codebase organized into logical modules
- **Easy Training**: Simple interface for fine-tuning models on custom datasets
- **Evaluation Metrics**: Built-in BLEU score evaluation for model assessment
- **HuggingFace Integration**: Seamless integration with HuggingFace Hub for model sharing
- **Flexible Configuration**: Centralized configuration system for easy customization
- **Version Control**: Semantic versioning system for tracking releases
## 🚀 Installation
### Prerequisites
- Python 3.8 or higher
- CUDA-capable GPU (recommended for training, optional for inference)
### Install from Source
1. Clone the repository:
```bash
git clone git@github-second:Galsenaicommunity/Wolof-NMT.git
cd Wolof-NMT
```
2. Install dependencies:
```bash
pip install -r requirements.txt
```
3. Configure environment variables:
```bash
# Copy the example environment file
cp .env.example .env
# Edit .env with your settings
# At minimum, you may want to set:
# - MODEL_CHECKPOINT (if using a custom trained model)
# - HF_TOKEN (if pushing models to HuggingFace Hub)
# - WANDB_API_KEY (if using Weights & Biases)
```
4. (Optional) Install as a package:
```bash
pip install -e .
```
## 🏃 Quick Start
### Configuration
Before using the translator, configure your environment variables. Create a `.env` file from the example:
```bash
cp .env …