doc-chatbot

Introduction to doc-chatbot: GPT x Pinecone x LangChain

The doc-chatbot project integrates cutting-edge technologies such as GPT, Pinecone, and LangChain to create an interactive platform for document-based conversations. It leverages these advanced technologies to enable dynamic discussions on a wide range of topics, making it versatile and applicable across various use cases.

Features of doc-chatbot

• Multiple Conversations: Users can set up multiple topics to talk about. This flexibility allows for extensive exploration of different subjects within one platform.

• File Handling: Users can upload any number of files per topic. The chatbot supports various formats, including .pdf, .docx, and .txt. These files are converted into embeddings and stored for future interactions.

• Chat Management: For each topic, users can create numerous chat windows, offering an organized way to manage multiple discussions simultaneously.

• Browser-Based Operations: Users can upload files, convert them into embeddings, store these embeddings, and delete Pinecone namespaces—all directly from the web browser without the need for additional software.

• Automatic Chat History Retrieval: The system automatically stores and retrieves chat history using local storage, ensuring users can effortlessly continue their discussions.

Technical Framework

The project is built using modern technologies:

• TypeScript: Ensures a type-safe environment for developing robust applications.
• Next.js: Provides a server-rendered React framework to help create scalable web applications.
• React: Delivers a component-based approach for building dynamic user interfaces.
• TailwindCSS: Offers utility-first styling, making it easy to create responsive designs.
• LangChain and Pinecone: Enhance the system's ability to process and store vector embeddings efficiently.

Setting up Locally

The doc-chatbot can be run locally by cloning the repository and following some setup steps:

1. Clone the Repository: Download the project using the command:

   git clone https://github.com/dissorial/doc-chatbot.git
   

2. Pinecone Setup: Create a Pinecone account, configure an index with specific settings, and generate an API key to enable the embedding functionality.

3. Install Packages: Use yarn install to install necessary packages.

4. Configuration: Set up the .env file to define the environment parameters.

5. Deploying and Running: Use npm run dev to start the application in development mode. Additional configurations are necessary for production deployment, including setting secrets for authentication.

Troubleshooting

• General Issues: Ensure you are using the latest Node version. If issues with documents arise, try converting them to text or checking for file integrity.

• Pinecone Related Issues: Confirm vector dimensions are set correctly. Be aware that free plan indexes are auto-deleted after inactivity and need to be refreshed to keep active.

Acknowledgments

The project is a derivative of a repository originally forked from GPT-4 & LangChain by mayooear, incorporating numerous enhancements and changes.

The doc-chatbot project provides a robust platform for document-focused interactions, harnessing the power of modern technologies to deliver an efficient and versatile solution for users looking to engage in meaningful conversations around diverse topics.