DocChat: Intelligent Document Q&A with Adjustable Response Style & Difficulty via Retrieval Augmented Generation (RAG)

About The Project

DocChat is a web application which uses Retrieval Augmented Generation (RAG) to answer user queries about a document given a desired response type and complexity.

The application utilizes:

• LangChain for chaining LLM operations
• OpenAI's text-embedding-ada-002 model for document embeddings
• Pinecone for vector storage and similarity search
• Streamlit for the user interface

Below are some of the key features of DocChat:

• Upload & Process PDFs: Users can upload their PDF files to be processed instantly
• Dynamic Querying: Ask questions related to document content with AI-driven responses
• Response Customization: Adjust response format and difficulty level
• Cloud-based Processing: Ensures quick and accurate retrieval of document insights.

Getting Started

Prerequisites: Python 3.8+, Pip, Git

1. Clone this repository and navigate to the local project folder. Activate your virtual environment (optional).
2. Install dependencies

   pip install -r requirements.txt
   

3. Create .env file in main directory

   touch .env
   

4. Set your environment variables

   OPENAI_API_KEY=your-openai-api-key
   PINECONE_API_KEY=your-pinecone-api-key
   INDEX_NAME=your-pinecone-environment
   

5. Run the app

   streamlit run app.py
   

Project View

About

Author: Ronoy Sarkar

Project Link: https://github.com/ronoys/DocChat

Acknowledgments

• README Template

(back to top)