HassnaaElshafei/Darija-to-Egyptian-Translation
Domaine:
natural language processingType de record:
software
Créateur:
Has
Hôte:
Moroccan Darija → Egyptian Arabic translation system with two inference pipelines: direct fine-tuned NLLB and a two-step AraT5 approach (Darija→MSA→Egyptian), wrapped in a Streamlit interface for interactive comparison.
# Darija to Egyptian Arabic Translation
An interactive Moroccan Darija to Egyptian Arabic translation system with two
inference pipelines, exposed through a Streamlit web interface for easy testing
and comparison.
---
## Live Demo (Streamlit)
Try the app here:
darija-to-egyptian-translat…
## Features
- Two translation approaches:
- Direct translation using a fine-tuned NLLB model
- Two-step translation using AraT5:
- Darija to Modern Standard Arabic (MSA)
- MSA to Egyptian Arabic using a LoRA-adapted AraT5v2 model
- Streamlit-based web interface
- Lightweight and modular design using LoRA adapters
- Fully based on open-source models
---
## Models Used
### Direct Translation (NLLB)
- Base model: facebook/nllb-200-distilled-1.3B
- Fine-tuned with LoRA for Darija to Egyptian Arabic
### Two-step Translation (AraT5)
- Darija to MSA: Saidtaoussi/AraT5_Darija_to_MSA
- MSA to Egyptian: UBC-NLP/AraT5v2-base-1024 with a LoRA adapter
This repository contains code only. Model adapters are downloaded automatically
from Hugging Face at runtime.
---
## Project Structure
.
├── app.py
├── smoke_test.py
├── requirements.txt
└── README.md
---
## Setup
Clone the repository and create a virtual environment:
git clone github.com / .git
cd
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
Install dependencies:
pip install -r requirements.txt
---
## Smoke Test (Optional)
Run the smoke test to verify that base models and adapters load correctly:
python smoke_test.py
---
## Run the Application
Start the Streamlit app:
streamlit run app.py
The application will open in your browser and allow you to enter Darija text,
select a translation method, and view the Egyptian Arabic output.
---
## Deployment
This project can be deployed using:
- Streamlit Community Cloud
- Hugging Face Spaces
---
## License
The code in this repository is released under the Apache 2.0 license.
Mod …