Forms and senses per the lexeme data model are sub entities of lexemes. This structure should be reflected in the sematic structure of the lexeme HTML view.
Currently the form/sense constitute 2nd level headlines, consequently individual forms/senses should each render a 3rd level headline - e.g. for forms, following the lemma logic, the list of representations separated by an i18n message.
Description
Description
Related Objects
Related Objects
Event Timeline
Comment Actions
As a result of this, the headings are the same size in the mobile view, e.g. on https://m.wikidata.org/wiki/Lexeme:L123