RAG-Supported AI-Agent Template

This repository serves as a template for building a Retrieval-Augmented Generation (RAG) AI-agent. It provides a modular structure to process PDF documents, split them into manageable text chunks, assign unique IDs, store them in a vector database using embeddings, and then query a language model (via Ollama) to generate context-aware responses.

---

🚀 Features

• PDF Document Processing: Extract text from PDF documents.
• Text Chunking: Split documents into smaller, overlapping chunks.
• Unique Chunk IDs: Generate unique identifiers for each chunk.
• Vector Database Storage: Use Chroma for efficient storage and retrieval of embeddings.
• RAG Pipeline: Retrieve relevant document chunks and generate responses using a language model.
• CLI Tools: Single-query and interactive multi-query command-line interfaces for querying.

---

📦 Required Packages

Install the necessary packages by running:

pip install ollama pypdf langchain-community chromadb

or

python -m pip install -r requirements.txt

🗂️ Project Structure

.
├── constants.py              # Configuration constants (MODEL, CHROMA_PATH, DATA_PATH, PROMPT_TEMPLATE)
├── functions
│   ├── database.py           # Functions to manage the Chroma database operations
│   ├── embedding.py          # Embedding function setup using OllamaEmbeddings
│   ├── id_calculation.py     # Functions for generating unique chunk IDs
│   ├── pdf_processing.py     # Functions for loading and splitting PDF documents
│   └── query.py              # Functions for querying the RAG pipeline
├── main.py                   # Main script with CLI implementations (run_single_cli & run_multi_cli)
└── README.md                 # This file

⚙️ Setup and Usage

1. Clone the Repository

git clone https://github.com/Make1tRain/RAG-template.git
cd RAG-template

2. Install the Required Packages

python -m pip install -r requirements.txt

3. Run setup.py for the creation of the DATA_PATH and CHROMA_PATH directories which are located inside constants.py

python setup.py

4. Prepare Your Data Place your PDF documents in the directory specified by the DATA_PATH constant in constants.py (default is ./data).

5. Running the AI-Agent Single Query CLI To run a single query against the RAG pipeline, execute:

python main.py "Your query here"

Interactive Multi-Query CLI
To launch an interactive session:

python main.py

Then type your queries. Type exit or bye to quit the session.

🔍 How It Works

Document Processing:

Loading: The agent loads PDF and CSV documents from the specified directory.

Splitting: Documents are split into smaller, overlapping chunks for easier processing.

Unique Chunk IDs: Each text chunk is assigned a unique ID based on its source, page number, and position, preventing duplicate entries in the database.

Database Management: Processed chunks are stored in a Chroma vector database. Only new, unique chunks are added, ensuring efficient updates.

Retrieval and Generation: The system retrieves contextually similar chunks based on the user's query using vector similarity, and then combines this context with the query to generate an answer using a language model.

📝 Template Purpose

Note: This repository is intended as a template to help you apply the RAG (Retrieval-Augmented Generation) approach. It is designed to be customized and extended based on your specific requirements, whether you are working with different data sources or integrating with other language models.

Note

• Also, this is a very basic template to be used and one I created to just learn how to implement RAG, so use at your own discretion.
• Finally, I used deepseek-r1:14b model for this template along with the given file structure for the chorma database and the data to be fed to the AI-Agent which can all be changed in constants.py