My guess: There is a ton of data online of just plain weights laying around. No human would ever actually read it. LLM's read it all during training.

▲

objektif an hour ago | parent [-]

But that would cover finite set of cases. I believe this result is more general.

	▲	rsalus 17 minutes ago \| parent [-]
		I mean it would make sense if the training using that data resulted in a general set of equations/patterns, it's still really surprising though...