Kenoubi comments on Fact Finding: Trying to Mechanistically Understanding Early MLPs (Post 3)