llama.cpp adds an initial ExecuTorch backend for edge and on-device deployment
Build b9951 of llama.cpp drops an initial ExecuTorch backend, planting a flag for GGUF-format models on PyTorch's on-device inference stack and the edge hardware it targets.