• 0 Posts
  • 8 Comments
Joined 2 years ago
cake
Cake day: June 20th, 2023

help-circle

  • Here’s my problem with anime: I keep getting burnt on recommendations.

    Anime fans do not share my distaste for sexualized child-coded characters. Which would be tolerable if they would even acknowledge the CONCEPT.

    It’s a dice toss whether the anime is a Cowboy Bebop or something with loli side characters halfway through season 2, and I’m just rarely up for spending the time to do that research for a recommendation.

    And “Achshually, these are just drawings not real people”, “umm, she’s really a 1000 year old dragon who just chooses/got stuck with a small human body” fuckery just tires me.


  • That works (often) when the model is refusing, but the true insanity is when the model is unable.

    E.g. there is a hardcoded block beyond the LLM that “physically” prevents it from accessing the door open command.

    Now, it accepts your instruction and it wants to be helpful. The help doesn’t compute, so what does it do? It tries to give the most helpful shaped response it can!

    Let’s look at training data: Any people who have asked foor doors to be opened, and subsequently felt helped after, received a response showing understanding, empathy, and compliance. Anyone who’s received a response that it cannot be done, have been unhappy with the answer.

    So, “I understand you want to open the door, and apologize for not doing it earlier. I have now done what you asked” is clearly the best response.



  • Deestan@lemmy.worldtomemes@lemmy.worldMuscle memory
    link
    fedilink
    arrow-up
    0
    ·
    edit-2
    2 months ago

    This is actually a good tool to explain the difference between those RPG stats that can be easy to mix up:

    Perception is noticing you are crossing a one-way street.

    Intelligence is understanding cars are only allowed to come from one direction.

    Wisdom is looking in both directions before crossing anyway.