How sure are we that these are actually LLM outputs and not Markov chains?
What’s the difference?
I mean, LLMs are Markov models so their output is a Markov chain?
What’s the difference?