RAG Model - Retrieval-Augmented Generation with Krypt Documentation

A Retrieval-Augmented Generation (RAG) system that processes and indexes documentation from the Krypt blockchain project, enabling intelligent document retrieval and question-answering capabilities.

Project Overview

This project implements a RAG pipeline that:

• Ingests multiple document formats (PDF, Markdown, CSV)
• Processes documents using LangChain and sentence transformers
• Stores embeddings in ChromaDB vector store
• Enables semantic search and retrieval-augmented generation

Features

• Multi-format Document Support: Process PDFs, Markdown, and text files
• Vector Embeddings: Uses sentence-transformers for semantic understanding
• Persistent Vector Store: ChromaDB for efficient similarity searches
• LLM Integration: Groq API integration for generative tasks
• Modular Architecture: Clean separation of concerns for extensibility

Tech Stack

• LangChain: Document processing and RAG orchestration
• ChromaDB: Vector database for embeddings storage
• Sentence Transformers: Semantic embedding generation
• Groq: LLM API for generative responses
• PyMuPDF & PyPDF: PDF document parsing
• FAISS: CPU-based similarity search

Project Structure

RAG_model/
├── data/
│   ├── notion/          # Notion exports (.md, .csv)
│   ├── pdf/             # PDF documents
│   ├── text_files/      # Text files
│   └── vector_store/    # ChromaDB persisted data
├── notebook/
│   ├── document.ipynb   # Main processing notebook
│   └── pdf_loader.ipynb # PDF loading utilities
├── src/                 # Source code modules
├── main.py              # Entry point
├── requirements.txt     # Project dependencies
└── README.md

Installation

1. Clone the repository

   git clone <repository-url>
   cd RAG_model

2. Create virtual environment (Python 3.13+)

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Set up environment variables

   cp .env.example .env
   # Add your Groq API key and other configuration

Usage

Running the RAG Pipeline

python main.py

Jupyter Notebooks

• document.ipynb: Main data processing pipeline
• pdf_loader.ipynb: PDF extraction and loading utilities

Document Processing

The system processes documents from data/ directories:

• PDFs from data/pdf/
• Notion exports from data/notion/
• Text files from data/text_files/

Vector embeddings are stored in data/vector_store/ using ChromaDB.

Configuration

Update .env file with:

GROQ_API_KEY=your_api_key_here

Dependencies

See requirements.txt:

• langchain & langchain ecosystem
• chromadb (vector database)
• sentence-transformers (embeddings)
• pymupdf & pypdf (PDF processing)
• faiss-cpu (similarity search)

Document Set

This project indexes the Krypt blockchain project documentation:

• React + Solidity decentralized application for Ethereum transactions
• Smart contract code and frontend implementation details
• Project report and technical specifications

Future Enhancements

• Support for additional document formats (DOCX, HTML)
• Real-time document updates
• Multi-language support
• Advanced query expansion techniques
• Hybrid search (semantic + keyword)
• Web UI for document exploration

License

MIT License

Author

Tejomai V