[energy123]

Using LLMs to clean those up is part of the workflow that you're responsible for (... for now). If you're hoping to get ideal results in a single inference, forget it.