← Back to context Comment by esafak 2 years ago They're different because Markov models are stateless whereas LLMs are stateful.https://en.wikipedia.org/wiki/Markov_property 1 comment esafak Reply Jensson 2 years ago Current LLMs are stateless as far as we know, their state when computing a new token is only the preceding text tokens, they don't store any metadata or save state from the previous calculations.
Jensson 2 years ago Current LLMs are stateless as far as we know, their state when computing a new token is only the preceding text tokens, they don't store any metadata or save state from the previous calculations.
Current LLMs are stateless as far as we know, their state when computing a new token is only the preceding text tokens, they don't store any metadata or save state from the previous calculations.