Logo Lanfrica

AhmedbBZ/AI-for-Tunisian-Heritage

Domaine:

natural language processing

Type de record:

software
Créateur:
Ahm
Hôte:
tunisian_rag_pipeline/ # Tunisian Heritage RAG Pipeline A production-ready Retrieval-Augmented Generation (RAG) system for question-answering over Tunisian heritage data, leveraging a local Large Language Model (LLM) for private, high-quality responses. ## 🏗️ Project Structure ``` ├── README.md # Project documentation ├── scrapper.py # Web scraper for data collection ├── Ai_Promptes_Caps/ # AI prompt capts ├── Architecture/ # Architecture documentation ├── tunisian_heritage_data/ # Heritage datasets │ ├── dataset_index.json │ ├── metadata/ # Metadata JSON files │ ├── pdfs/ # PDF documents │ ├── raw_html/ # Raw HTML files │ └── texts/ # Text documents ├── tunisian_rag_pipeline/ # Main RAG pipeline │ ├── chat.py # Interactive chat interface │ ├── requirements.txt # Python dependencies │ ├── test_retrieval.py # Retrieval tests │ ├── config/ # Configuration files │ ├── scripts/ # Utility scripts (build, query, fine-tune, diagnose) │ ├── src/ # Source code │ │ ├── data/ # Data processing (chunking, ingestion, preprocessing) │ │ ├── embeddings/ # Embedding models │ │ ├── llm/ # LLM generation and prompts │ │ ├── pipeline/ # RAG pipeline and intent detection │ │ ├── retrieval/ # Vector store and retriever │ │ └── utils/ # Helper utilities │ ├── tests/ # Unit tests │ └── vector_db/ # ChromaDB persistent storage └── vector_db/ # Alternative vector database location ``` ## 🚀 Quick Start(first u shood download an LLM Localy ) ### 1. Install Dependencies ```bash pip install -r requirements.txt ``` ### 2. Build the Vector Database ```bash python scripts/build_vector_db.py ``` ### 3. Query the System ```bash python scripts/query.py "What caused the Tunisian revolution?" ``` ### 4. Interactive Chat ```bash python chat …