llama.cpp

71a062c5 - add gguf_get_tensor_nbytes (factored out ggml_nbytes_impl & ggml_fill_nb helpers)