@mira Thanks! These seem interesting but not quite what I was asking for :-) (though you probably weren't optimising for that anyway)
@rune Ah, that's what you meant. Seems true 👍
@mira
Makes sense, hadn't connected regularization with simplicity.
(tho I don't think I understand regularization enough yet to understand why it'd result in simplicity).
@rune I think that we mostly can't reach into the resulting buckets of neural nodes and change stuff we don't like.
@Azure Non-power users were a mistake
@WomanCorn I wonder how much advantage LLMs give someone with search-engine experience on a search engine in 2017.
E.g. ChatGPT lied to me about linearity in Jacobians, which is often the kind of thing I want to discover.
Probably still enough advantage.
I operate by Crocker's rules[1].