Zap11/llama-cpp-python-wheels-LMF2.5

1# llama-cpp-python — Free-Tier Friendly Wheel2 3This Space provides a **prebuilt `llama-cpp-python` wheel** designed to work4**reliably on Hugging Face Free tier Spaces**.5 6No compilation. No system packages. No build failures.7 8If your Space crashes during `pip install llama-cpp-python`, this wheel is the fix.9 10---11 12## Optimized for Hugging Face Free Tier13 14Hugging Face Free tier Spaces are:15 16- CPU-only17- Limited in memory18- Not suitable for native compilation19 20This wheel is built **ahead of time** so it can be installed instantly without21triggering CMake, compilers, or BLAS detection.22 23---24 25## What this wheel gives you26 27- ✅ Works on **HF Free tier CPU Spaces**28- ✅ Linux (ubuntu-22.04 compatible)29- ✅ Python 3.1030- ✅ OpenBLAS enabled (`GGML_BLAS=ON`)31- ✅ No system dependencies required32- ✅ No build step during Space startup33- ✅ Fast, reliable `pip install`34 35---36 37## How to use in a Space (Free tier)38 391. Download the wheel from the GitHub repository402. Upload it to your Space413. Install it in your Space startup:42 43 44 45pip install llama_cpp_python-*.whl>46 47 48## That’s it — your Space will start without build errors.49 50## Build details51 52This wheel was built using:53abetlen/llama-cpp-python (recursive submodules)54OpenBLAS (GGML_VENDOR=OpenBLAS)55scikit-build-core56ninja57python -m build --wheel --no-isolation58 59## Build environment:60 61OS: Ubuntu 22.0462Python: 3.1063 64## Why not build from source on HF?65 66On Free tier Spaces, building from source often fails due to:67Missing compilers68Missing BLAS libraries69Memory limits70Build timeouts71This prebuilt wheel avoids all of those issues.72 73## Notes74 75CPU-only (no CUDA)76Intended for inference workloads77Not an official upstream release78 79## Credits80 81All credit goes to the maintainers of llama-cpp-python and llama.cpp.82This Space exists solely to make Free tier usage painless.