dboa9/translation-service

Darija translation-service # Darija Translation Service A translation service for Moroccan Darija (Arabic dialect) to English and vice versa, using various machine learning models and datasets. ## Features - Multiple translation models support - Dataset management and validation - Web interface for translation - Comprehensive testing suite - EC2 deployment support ## Project Structure ``` ├── core/ # Core functionality │ ├── interfaces/ # User interfaces │ ├── translation/ # Translation services │ ├── dataset/ # Dataset handling │ └── validation/ # Validation utilities ├── config/ # Configuration files ├── tests/ # Test suites ├── deployment/ # Deployment scripts └── frontend/ # Web interface ``` ## Setup 1. Install dependencies: ```bash pip install -r requirements.txt ``` 2. Configure environment: - Edit `config/credentials.py` with your Hugging Face API token - Configure model settings in `config/model_config.yaml` 3. Run tests: ```bash ./run_tests.sh ``` 4. Start the interface: ```bash ./run_interface.sh ``` ## Models The service supports multiple translation models: - AnasAber/seamless-darija-eng - Helsinki-translation-English_Moroccan-Arabic - And more... Models are downloaded on-demand and cached locally. ## Development - Python 3.8+ - Uses Git LFS for large file handling - Comprehensive test suite in `/tests` - EC2 deployment scripts in `/deployment` ## License MIT License