README.md

PDF QnA Bot

This repository contains a Streamlit application that uses OpenAI's GPT-3.5-turbo model to answer questions based on the content of PDF documents. The application is containerized using Docker and uses Conda for dependency management.

Features

• Extract text from PDF files
• Chunk text for processing
• Create a vector store using embeddings
• Use OpenAI GPT-3.5-turbo for question-answering
• Display answers in a Streamlit web application

Prerequisites

• Docker installed on your machine
• OpenAI API key

Project Structure

.
├── Dockerfile
├── README.md
├── environment.yml
├── .env
├── main.py
├── multidocs_palm.py
├── htmlTemplates.py
└── resources
    ├── image1.png
    └── image2.png

Environment Variables

Create a .env file in the root directory of the project and add your OpenAI API key:

OPENAI_API_KEY=your_openai_api_key

Replace your_openai_api_key with your actual OpenAI API key.

Installation

Build the Docker Image

docker build -t streamlit-conda-app .

Run the Docker Container

docker run -p 8501:8501 --env-file .env streamlit-conda-app

Access the Application

Open your browser and navigate to http://localhost:8501 to access the Streamlit application.

Usage

1. Upload PDFs: Click on the "Choose a PDF file" button to upload one or more PDF files.
2. Ask Questions: Enter your questions in the text input box.
3. Get Answers: The bot will process the PDFs and provide answers based on the content.

Images

Here are some images showing the bot in action:

Bot Working - Step 1

Bot Working - Step 2

Contributing

Feel free to open issues or submit pull requests with improvements.

License

This project is licensed under the MIT License. See the LICENSE file for details.