Use model->gguf_kv for loading the template instead of using the C API.#10868

slaren merged 2 commits intoggml-org:masterggml-org/llama.cpp:masterfrom

dranger003:masterdranger003/llama.cpp:masterCopy head branch name to clipboard