Loading
ollama: don't build llama.cpp twice
buildGoModules runs preBuild hooks in the module download fixed-output derivation, so llama.cpp got built and immediately discarded, to then immediately be built again for the actual Go build.
buildGoModules runs preBuild hooks in the module download fixed-output derivation, so llama.cpp got built and immediately discarded, to then immediately be built again for the actual Go build.