AI Arabic Grammar Parser
Mu'rib | مُعرب
Mu'rib | مُعرب
Sdaia Allam Challenge Project
Project Overview
Revolutionizing Arabic grammar learning through Mu'rib: A Python-based application powered by a manually curated dataset and advanced Prompt Engineering to achieve high-fidelity syntactic dependency parsing.
- Developed a mobile application using Python and KivyMD, allowing users to input Arabic sentences and receive accurate grammatical parses.
- Built and fine-tuned the 3lam (13B) model with a manually curated dataset of 1,292 sentences covering 22 syntactic dependency types.
- Integrated the model with a Retrieval-Augmented Generation (RAG) system and applied prompt engineering to enhance parsing accuracy and context understanding.
Core Competencies & Skills
Programming & Frameworks
Python, Kivy, KivyMD
Large Language Models (LLMs)
3lam (13B), Model Fine-tuning
Natural Language Processing
Syntactic Parsing, Dependency Grammar
Advanced AI Techniques
RAG, Prompt Engineering
Data Science & Engineering
Dataset Curation, Labeling, Structuring
AI Implementation
Model Optimization, Context Understanding
Software Architecture
LLM Integration into Mobile Apps
Project Output Sample
Project Presentation
Detailed presentation covering the project lifecycle and technical depth.
GitHub Repository
Access the source code and implementation details on GitHub.