Comment by brcmthrowaway

What is the secret sauce that makes LLM better than a Markov chain?