• 0 Posts
  • 45 Comments
Joined 1 year ago
cake
Cake day: June 12th, 2023

help-circle









  • I’m skeptical that an LLM could answer questions as effectively just with documentation. A big part of the value in stack overflow and similar sites is that the answers provided come from people who have experience with a given technology and have some understanding of the pain points. Often times you can ask the wrong question and still get a useful answer because the context is enough for others to figure out what you might be confused by.

    I’m not sure an LLM could do the same just given the docs, but it would be interesting to see how close it could get.



  • Idk if that would be a good business decision. They would want it to be free and easy to start a channel still, so it would mean once your channel gets to a certain popularity google makes the deal progressively worse. This would create a big incentive for competition if all your biggest content creators are suddenly paying over cost to subsidize smaller channels.

    Not that this would be a bad thing, but I don’t see why google would ever want to risk it.



  • Asifall@lemmy.worldtoTechnology@lemmy.world...
    link
    fedilink
    English
    arrow-up
    21
    ·
    9 months ago

    The mentioned but unsupported link to “general intelligence” reeks of bullshit to me. I don’t doubt a modified LLM (maybe an unmodified one as well) can beat lossless compression algorithms, but I doubt that’s very useful or impressive when you account for the model size and speed.

    If you allow the model to be really huge in comparison to the input data it’s hard to prove you haven’t just memorized the training set.