Project: chat-with-docs

Chat with Your Docs

Demo

In this demo, a PDF file from the demo folder is uploaded, and questions are asked about the content of the PDF. The chatbot provides answers to the questions based on the PDF content.

Welcome to chat-with-docs!

This project empowers users to have interactive, conversational interactions with multiple documents at once, leveraging the capabilities of GPT and LangChain.

Table of Contents

• Features
• Technologies
• Setup and Installation
  • Prerequisites
  • Frontend
  • API (Backend)
• Usage
• Demo

Features

• Document Conversations: Chat with multiple documents simultaneously for a comprehensive understanding of your content.
• AI-Powered Chat: Leverages GPT for sophisticated natural language processing and conversational AI.
• LangChain Integration: Enhances interactions with your documents by providing context and continuity in conversations.
• Supports Multiple File Formats: Upload and chat with documents in PDF, DOCX, and TXT formats.
• Streaming Support: Engage in real-time conversations with streaming responses for a more interactive chat experience.
• Chat History: Maintains chat history for each conversation, allowing you to refer back to previous interactions and continue where you left off.
• Vector Database: Utilizes Pinecone for efficient indexing and retrieval of document information.
• FastAPI Backend: Fast and reliable server-side API with seamless performance.
• React Frontend: Modern, responsive, and user-friendly interface for interacting with your documents.

Technologies

• Python: Used for building the backend of the application.
• FastAPI: The web framework used for the backend API, offering high performance and asynchronous capabilities.
• LangChain: Integrates large language models (LLMs) and vector databases for contextual conversational experiences.
• React: Powers the modern and interactive frontend user interface.
• Pinecone: Utilized as the vector database for efficient indexing and retrieval of document information.
• Ada Embedding Model: Used for embedding documents into vector space for efficient retrieval and contextual conversations.
• GPT LLM: Provides powerful language understanding and generation capabilities for the chatbot.

Setup and Installation

Prerequisites

• OpenAI API Key: You will need to create an API key in OpenAI to access GPT and Ada embedding models. Create a .env file in the server/ directory and add the API key as OPENAI_API_KEY (refer to the provided .env.example file).

• Pinecone API Key and Index: This project uses Pinecone as the vector database. Register here and create an index. Add the API key and index to the .env file in the server/ directory as PINECONE_API_KEY and PINECONE_INDEX.

Frontend

To set up the frontend of the project, follow the instructions provided in client/README.md.

API (Backend)

For the API (backend), refer to the instructions outlined in server/README.md.

Usage

Once you've set up the project, use it as follows:

• Start the Backend Server: Run the FastAPI server to initialize the backend.
• Launch the Frontend: Start the React development server to use the application.
• Upload and Chat: Upload your documents in PDF, DOCX, or TXT format and start conversing with them naturally and intuitively. Enjoy real-time streaming responses for an enhanced chat experience.