llama.cpp

ggml, server: add ggml_backend_dev_reset() for sleep mode

#25271

Open