the claim i'm making is that a lot of it is in the weights, yeah.

i dont think this has to contradict the platonic thesis, although i also am not sure what that is exactly and maybe it does contradict the strong version. and a big reason for this is because 'personalities' are way more complex than platonic forms. models can share platonic primitives but still have personalities (which are the results of a huge accumulation of memories, adaptations, and differences in architecture) that are non-transferrable