llama.cpp
llama.cpp copied to clipboard
Add precompiled files for windows x64 and python script instead of bash script.
Precompiled files for windows x64 via Cmake.
Please, do not distribute executable binaries in repository.
binaries should be in Releases, not repo.
Perhaps we could expose the artifacts generated in the CI pipeline.
And eventually make a version of the project to which the releases would be related to the binaries.
It should be built only my core maintainers. Not by fresh github account with shady name.

Why doesn't the Github Actions workflow build file publish the build result files as github artifacts? That'd allow downloading the build result of every action run.
Not by fresh github account with shady name.
Lol, why you are so rude? Yes Im new here, but that is not concern to blame me...I simply dont know how to do it right. You can tell me in normal words without agresion.
Haven't had time to put do this right (i.e. make a PR and have the workflow create a release with multiple artifacts), but if someone just wants the Windows binaries, here I have them being built with GH actions with a minimal fork: https://github.com/Hades32/llama.cpp/actions/runs/4441656002
In any case, this project has basically zero dependencies. Just clone this repo and type make
and you'll have your binaries in under 10 seconds.
Closing since we don't want to have unreproducible executables committed in the git.