Check the git repo for llama.cpp.

Model weights in torrent magnet:?xt=urn:btih:b8287ebfa04f879b048d4d4404108cf3e8014352

Runs slowly but without trouble on an intel i7 with 32 gb ram.