ggml

Why we included this project

If you have run a local LLM or transcription model on a laptop, you have probably used ggml without knowing it: it is the low-level C tensor library behind llama.cpp and whisper.cpp. It handles tensor operations, automatic differentiation, and integer quantization, which is what makes large models practical on ordinary hardware, and it does so with no third-party dependencies and no runtime memory allocations. Developers building their own inference engines or edge runtimes will find it a compact foundation rather than a turnkey application. Because the project is under active development and much of the work happens in the llama.cpp and whisper.cpp repos, treat it as a building block to integrate and extend. Reading its source is also a good way to understand how quantized inference actually works.