The intent is the same, but as I understand it LLMs don't tokenize based on lemmas, though some of the tokens probably line up with them.