ollama.sh: installs from official repos, enables ollama.service, notes
GPU sharing caveat with llama.cpp. For NVIDIA/AMD GPU variants use
ollama-cuda or ollama-rocm from AUR instead.
llama-cpp.sh: standalone inference CLI and server via yay (covers both
official repos and AUR). Both modules coexist at the package level;
docker/podman/cockpit modules confirmed conflict-free (all use --needed,
podman+cockpit base packages already in core-packages.sh).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The_miro