Chat With Documents

Open it Here!

Open it Here

Overview

Chat With Documents is a Streamlit application that leverages Retrieval-Augmented Generation (RAG) to enable interactive, context-aware conversations with large language models (LLMs). Users can upload documents or provide URLs, and the app indexes the content using a vector store (Chroma) to supply relevant context during chats. The app supports multiple LLM providers including OpenAI, GROQ, and Anthropic.

Features

• Interactive Chat Interface: Engage in a conversation with an LLM enhanced with contextual information from your documents.
• Document Upload: Upload files in PDF, TXT, DOCX, or Markdown formats.
• URL Integration: Extract and index content directly from webpages.
• RAG (Retrieval-Augmented Generation): Combines LLM responses with relevant document context.
• Multiple LLM Providers: Choose between OpenAI, GROQ, and Anthropic models.
• Streaming Responses: Experience real-time streaming responses for a fluid chat experience.
• Vector Store Integration: Uses Chroma for indexing and retrieving document embeddings.

Demo

Check out the live deployed app: Chat With Documents on Streamlit

Installation

Prerequisites

• Python 3.8+ (tested with Python 3.11)
• Git

Setup Steps

1. Clone the Repository:

   git clone https://github.com/billy-enrizky/chat-with-documents.git
   cd chat-with-documents

2. Create and Activate a Virtual Environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the Required Dependencies:

   pip install -r requirements.txt

4. Configure Environment Variables:

   OPENAI_API_KEY=your_openai_api_key_here
   GROQ_API_KEY=your_groq_api_key_here
   ANTHROPIC_API_KEY=your_anthropic_api_key_here

   you can enter these keys via the sidebar in the app.

Usage

1. Run the Application:

   streamlit run streamlit_app.py

2. Interact with the App:

   • Configure the LLM Provider: Use the sidebar to select your preferred LLM provider (OpenAI, GROQ, or Anthropic) and model.
   • Upload Documents or Enter URLs: Provide documents (PDF, TXT, DOCX, MD) or URLs to supply context.
   • Chat: Type your message in the chat input. Depending on your settings, the app will use RAG (if enabled and documents are loaded) or a standard LLM chat to respond.
   • Toggle RAG: The "Use RAG" toggle is available in the sidebar once a vector database is initialized with your documents.

Project Structure

chat-with-documents/
├── README.md              # This file
├── requirements.txt       # Python dependencies
├── streamlit_app.py       # Main Streamlit application
└── rag_utils.py           # Utility functions for RAG (indexing, document loading, etc.)

Troubleshooting

Common Issues

• pysqlite3.dbapi2.OperationalError: no such table: collections
  This error indicates that the Chroma vector store is not correctly initialized. It can occur if an old persisted database is used. To resolve:

  • Delete the Persisted Directory: Modify the initialize_vector_db function to remove any existing directory before initializing a new database.
  • Run Migration: Alternatively, follow Chroma’s migration guide to update the schema.

• ValueError: Expected Embedings to be non-empty list or numpy array, got [] in upsert.
  This error occurs if empty document chunks are passed to the embedder. Ensure that document splitting filters out empty or whitespace-only chunks.

• Torch Module Path Error:
  If you see errors like:

  RuntimeError: Tried to instantiate class '__path__._path', but it does not exist!
  

  it is due to Streamlit’s local sources watcher inspecting Torch’s custom modules. A monkey-patch can be applied at the very start of your script to safely bypass these errors.

Contributing

Contributions are welcome! If you’d like to contribute:

1. Fork the repository.
2. Create a new branch (git checkout -b feature/YourFeature).
3. Make your changes and commit them (git commit -m 'Add some feature').
4. Push your branch (git push origin feature/YourFeature).
5. Open a pull request.

For major changes, please open an issue first to discuss what you would like to change.

License

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

Contact

For any questions or feedback, please open an issue on the GitHub repository or contact the repository owner.

---

Enjoy chatting with your documents!