llama.cpp

llama : add missing call to ggml_backend_load_all()

#22752

Merged