SNS Chat Application

SNS Chat is an RAG-powered chatbot application designed to assist employees at SNS Network with their queries. It leverages LangChain, NVIDIA, and HuggingFace APIs to provide accurate and efficient responses.

Features

• AI Chat: Utilise powerful language models to provide accurate answers to employee questions.
• Multi-query Retrieval: Generate multiple perspectives on user questions to improve retrieval accuracy.
• Contextual Compression: Compress retrieved documents with reranking for efficient information extraction.
• Flask Web Interface: A simple and interactive web interface built using Flask.

Installation

1. Clone the repository:

   git clone <repository_url>
   cd sns_chat_v2

2. Install dependencies:

   pip install --upgrade pip
   pip install --no-cache-dir -r requirements.txt

3. Set environment variables:

   export NIM_IP='YOUR_IP'
   export PHOENIX_IP='YOUR_IP'  # Comment out if not using Phoenix
   export LLM_MODEL='YOUR_MODEL'  # Example: mistralai/Mistral-7B-Instruct-v0.3

Usage

1. Run the Flask application:

   waitress-serve --listen=*:5000 sns_chat_v2:app

   gunicorn --bind 0.0.0.0:5000 sns_chat_v2:app

2. Access the application: Open your web browser and navigate to http://localhost:5000.

3. Interact with the chatbot: Enter your query in the provided input field and submit it. The chatbot will respond with the relevant information.

Docker

You can also run the application using Docker:

1. Pull and run the Docker image:

   docker run --rm -it -e NIM_IP=YOUR_IP -e PHOENIX_IP=YOUR_IP -p 5000:5000 --name sns_chat mulberrydunes/sns_chat:0.0.5

Clearing Chat History

To clear the chat history, click on the "Clear History" button on the web interface. This will reset the chat history.