That's impressive for an LLM. I've rarely had any make sense for me. As long as it can accidentally do a few basic bits of logic, it's good enough for me to add to my toolkit, this is a great usecase.