QA Bot with Retrieval-Augmented Generation (RAG)

This repository contains the implementation of a Question Answering (QA) Bot powered by a Retrieval-Augmented Generation (RAG) model. The bot uses a vector database for efficient document retrieval and a generative model for producing coherent answers based on the content of the documents.

Table of Contents

• Introduction
• Features
• Requirements
• Installation
• Usage
• Running the Application
• Docker Setup
• Future Enhancements

Introduction

This project demonstrates a QA system that answers user queries based on document content. It integrates a RAG model that uses document embeddings for retrieval and a generative model (like Cohere API) to generate answers.

The system includes:

• A backend that manages document embedding and retrieval using Pinecone DB.
• A frontend interface built using Streamlit for real-time user interaction.

Features

• Upload PDF documents and extract content.
• Store and retrieve document embeddings using a vector database.
• Generate coherent, contextually relevant answers using a generative model.
• Real-time query answering with the ability to view the relevant document segments.

Requirements

To run this project locally, you need the following dependencies:

• Python 3.8 or higher
• Streamlit for the frontend
• Pinecone for document retrieval
• Cohere API (or an alternative generative model)
• Docker (optional, for containerization)
• Additional Python libraries (see requirements.txt)

Installation

Follow these steps to set up the QA bot on your local machine:

1. Clone the repository:

   git clone https://github.com/vaibhv02/GenAI_Project.git
   cd GenAI_Project

2. Install the required dependencies: It's recommended to create a virtual environment first:

   python3 -m venv myenv
   source venv/bin/activate  # for Windows: venv\Scripts\activate

   Then, install the dependencies:

   pip install -r requirements.txt

3. Configure API Keys: You'll need API keys for Pinecone and Cohere (or the generative model of your choice). Create a .env file in the root directory with the following content:

   PINECONE_API_KEY=your-pinecone-api-key
   COHERE_API_KEY=your-cohere-api-key

Usage

Uploading Documents and Asking Questions

1. Once the application is running, you can upload a PDF document from the frontend interface.
2. After uploading the document, ask a question related to its content.
3. The system will retrieve relevant information from the document and generate a response using the RAG model.

Running the Application

To run the QA bot on your local machine:

streamlit run Dashboard/app.py

Access the app by navigating to http://localhost:8501 in your browser. Upload a document (PDF), type a question, and receive an answer based on the document content.

Docker Setup

If you prefer to use Docker for deployment, follow these steps:

1. Build the Docker image:

   docker build -t genai .

2. Run the container:

   docker run -p 8501:8501 genai

Open your browser and go to http://localhost:8501 to use the application.

Future Enhancements

• Improved Query Processing: Enhance the query handling mechanism to improve retrieval accuracy.
• Multi-User Support: Scale the application to support multiple users concurrently.
• Additional Document Formats: Extend support to additional file formats (e.g., Word documents, plain text files).