The key unresolved question is: why do LLMs follow instructions, especially system prompt?And I don’t see a good answer which is not worrying if not spooky.
they are quite unreliable at consistently following instructions, it's one part training, one part inherent (imu)
What’s unresolved about it?They are assembled to. We tweak the numbers, bit by bit, trillions of times, to maximize the chances of desirable output.
strideashort · · focus · HN ↗
And I don’t see a good answer which is not worrying if not spooky.
verdverm · · focus · HN ↗
dd8601fn · · focus · HN ↗
They are assembled to. We tweak the numbers, bit by bit, trillions of times, to maximize the chances of desirable output.