this is about file formats / data formats _and_ actually how people write i think.
its a bit weird to mix those concepts (sry edits was confused).
file formats just need parsers if you can extract the text its a non issue ? it might help reduce cpu cycles if u dont use PDF or complex file formats thats fair.
on changing how we actually wrtie:
answer would be no in that case, an LLMs primary purpose is to understand our human language. so if it would perform bad its a program issue not a human language issue. a paper should include data it is based on ideally, so maybe not omit that.
this is about file formats / data formats _and_ actually how people write i think.
its a bit weird to mix those concepts (sry edits was confused).
file formats just need parsers if you can extract the text its a non issue ? it might help reduce cpu cycles if u dont use PDF or complex file formats thats fair.
on changing how we actually wrtie: answer would be no in that case, an LLMs primary purpose is to understand our human language. so if it would perform bad its a program issue not a human language issue. a paper should include data it is based on ideally, so maybe not omit that.