streamlit-llm-chatbot

A simple chatbot using Groq backed LLM using Retrieval Augmented Generation (RAG) with Langchain.

It is a streamlit app that lets you make retrieval augmented generation from a web source using LLM and LangChain. It is supported by Groq API to make the LLM inference.(It is free for personal use with limits, highly recommended to try it out!)

What can be done?

1. Connect to a Llama3 or Gemma using your Groq API key, and chat with it!
2. Input any website, and query/chat based on questions from the website and LLM's knowledge.
3. Input a Git repository, and query/chat based on the files in the repository.

Try out the app on Streamlit - streamlit-llm-chatbot

Setup instructions

Create a virtual environment to install from requirements file. You need Python 3.10 or higher.

pip install -r requirements.txt

Run

To run the app on streamlit, use the following command:

streamlit run main.py

The application will automatically open in your default browser. You can check the logs in the terminal you spun the streamlit server.

LLMs and embeddings supported

The available LLM backends as supported by Groq are as follows:

• Llama3: "llama3-70b-8192" - Enabled by default
• Llama3: "llama3-8b-8192"
• Gemma: "gemma-7b-it"

Embeddings: Supports Hugging Face Embeddings by default, which is now hardcoded to "sentence-transformers/all-mpnet-base-v2". Can be dynamic in the future.

Credits

A big shoutout to all the detailed documentations from Langchain and Streamlit to help build the app. Groq has been generous with their rate limits for the LLMs which enabled the development of the app. Thanks to pip-chill for a chill pip freeze. Used Loom for the demo video of the app.