Lexical analysis models can help in tokenisation process, could they potentially help the LLM in interpretation of the commands?