- https://github.com/Atome-FE/llama-node is quite out of date - doesn't support recent/current llama.cpp functionality
--n-gpu-layers
--gpu-layers
fixed a typo in the MacOS Metal run doco