# LLama Herder

- manages multiple llama.cpp instances in the background
- keeps track of used & available video & cpu memory
- starts/stops llama.cpp instances as needed, to ensure memory limit is never reached