StanislavKrym comments on [Paper] Does Self-Evaluation Enable Wireheading in Language Models?