Comments by "Chrysippus" (@4.0.4) on "DeepSeek explained - The Chinese AI that shocked the world" video.

  1. Biggest surprise to me is that the thinking and reflection, wasn't trained. It just emerged from training. Only accurate answers and presence of <thinking> were rewarded!
    77
  2. Honestly, the 70B is probably not too far behind the 671B. For example llama 405B isn't several times smarter than llama 70B.
    11
  3.  @chad_usa  just a regular home computer, I see.
    6
  4.  @darekmistrz4364  only accurate responses were rewarded, not all the anthropomorphic reflection during thinking. That's emergent.
    1