Distribute and run LLMs with a single file.
C++ cross-platformggufllama-cpplocal-ailocal-inferencelocal-llmopen-source-aisingle-file-executablespeech-to-text
description: Build llamafile using the cosmocc toolchain
Build Llamafile
Build the project using the Cosmopolitan toolchain.
First, ensure the toolchain is available:
if [ ! -d .cosmocc/4.0.2 ]; then
build/download-cosmocc.sh .cosmocc/4.0.2 4.0.2 85b8c37a406d862e656ad4ec14be9f6ce474c1b436b9615e91a55208aced3f44
fi
Then build:
.cosmocc/4.0.2/bin/make -j $(nproc)
Adapt nproc to the OS where you are building, (e.g. sysctl -n hw.physicalcpu on mac)
Build outputs will be in o/$(MODE)/.