ggml tensor library repo listed on GitHub
First seen Nov 16 Updated Nov 16, 11:44 UTC Heat 18 medium evidence
The ggml-org/ggml repository, a tensor library for machine learning, appeared on GitHub trending. It is written in plain C/C++ with no dependencies and supports many CPU architectures and backends. Core changes should go through llama.cpp.
- What to do
- Open core library PRs in llama.cpp, as the README instructs.
- Who should care
- Developers building local or embedded inference engines.
Why it matters: The MIT-licensed C/C++ library has no dependencies, so it can be embedded for local inference across CPU, GPU, NPU and browser targets.
Key facts
- License
- MIT
- Quantization
- 2- to 8-bit integer, plus MXFP4 and NVFP4
- Targets
- x86, ARM, RISC-V, LoongArch, PowerPC, s390x, WebAssembly
Timeline
Coverage
-
ggml-org/ggml: Tensor library for machine learninggithub.com · GitHub · 16k stars
Evidence
Medium evidence: several community reports point the same way, but there is no official confirmation yet.
Details written Nov 16, 06:23 UTC from the sources listed above.