"Evaluating results" is completely beyond the ken of a model which will blithely make results up, bullshit explanations for them, and if pressed invent the citations underpinning the expectations. It's spectacularly "eat a small rock every day" unfit to evaluate results