Make your Raspberry Pi, into a pet rock. That can see.

You need poetry.

Run: poetry install

To download the requisite models, run: poetry run ./download_models.py

Then, to run the server: poetry run ./server.sh

This starts screen with both flask and llama_cpp.server running. You can access the server at port 5050 - localhost:5050 if running locally.

First, you need poetry and screen: sudo apt install python3-poetry screen

Then install as normal.

Sometimes you'll run into a DBus error when trying to use poetry. Do this, and source ~/.bashrc, and things should work properly. echo "export PYTHON_KEYRING_BACKEND=keyring.backends.fail.Keyring" >> ~/.bashrc

Petrock is run through a Flask app that communicates with llama_cpp.server using the configuration found under petrock/server_config.json.

This server serves the Moondream2 VLM and a quantized LLama3 LLM.