Comment by wfn

> but then the shell commands were actually running llama.cpp, a mistake probably no human would make.

But in the docs I see things like

    cp llama.cpp/build/bin/llama-* llama.cpp

Wouldn't this explain that? (Didn't look too deep)