I don‘t think so. My impression with LLMs is that they correct typos well. I would imagine this happens in early layers without much impact on the remaining com

I don‘t think so. My impression with LLMs is that they correct typos well. I would imagine this happens in early layers without much impact on the remaining computation.