⬢github C++ · 15.8K ★ +140 since we first saw it · pushed 1 d ago · MIT
ggml-org/ggml
Tensor library for machine learning
ggml is a lightweight, dependency-free C/C++ tensor library for machine learning. It runs on many CPU architectures plus GPU, NPU, and browser backends, with SIMD-optimized kernels, 2-8 bit integer quantization (including MXFP4/NVFP4), and zero runtime memory allocations. It powers llama.cpp and can be built with a simple CMake setup.
Why now: The project just released its v1.0.0 milestone version, which drew attention on Hacker News.
Who it is for: ML developers building efficient local inference or embedding tensor computation into portable, cross-platform applications.
machine-learning
Stars over our 51 snapshots: 15.6K to 15.8K, since 12 h ago.
Where people talked about it
- Yhn Ggml v1.0.0 8 min ago
API: https://socialmediatrends-api.osmike.com/v1/repos/ggml-org/ggml