4 ms·

I think that "LLMs can never X" is just always false.

• theWreckluse 2y ago
  "LLMs can never predict the next word"
• solomatov 2y ago
  LLM can never solve a halting problem (because no one can using a Turing machine).
  • woctordho 2y ago
    A finite-size LLM can solve the finite-size halting problem, and an infinite-size LLM can solve the infinite-size halting problem
    • solomatov 2y ago
      Halting problem input has finite size (i.e. it’s a Turing machine)