Simple Docker Compose to load gpt4all (Llama.cpp) as an API and chatbot-ui for the web interface. This mimics OpenAI's ChatGPT but as a local instance (offline).

git clone https://github.com/mkellerman/gpt4all-ui.git
cd gpt4all-ui

Models should be copied to models folder in the root of the gpt4all-ui folder.

If you have your models located somewhere else, you can create a alias (link) to your folder.

ln -s /Users/Me/Documents/Models models

docker-compose up -d

And open http://localhost:3000

chatboi-ui (web): https://github.com/mckaywrigley/chatbot-ui

llama-cpp-python (api): https://github.com/abetlen/llama-cpp-python