Thoughts on benchmarking of function modeling