π©
ggml is the foundational C tensor library behind llama.cpp and whisper.cpp, enabling efficient LLM inference with integer quantization and broad hardware support. It runs large language models locally with zero runtime memory allocations and no third-party dependencies.
