whether they count as the same object or hypothetical objects seems like a matter of degree/interpretation. and transformers even have a guarantee that human brains dont have: that activation vectors from previous "moments" are added to the current activations without modification. and maybe the layers are more different than you and your previous self, but how different? what if you just duplicated the layers in a transformer, would that fix the problem you're pointing at?
same thread: 2005114377255748082 2005115563102928968 2005737374622572892 2005739702029279445 2005742132854939992 2005742859601993809 2005744231797952840 2005747421935198271 2005749837187407981 2005752910450401754 2005757046810108155 2005757704644759639 2005772140004680060 2005772894698352781 2005776285113704952 2005792023303864830 2005799686834381285 2005810760249647421 2005816839180329424 2005820992560730474
Reproduced against link rot, credited and linked to its original. Yours and you’d rather it weren’t here? Open an issue.