• 0 Posts
  • 6 Comments
Joined 2 years ago
cake
Cake day: July 11th, 2023

help-circle

  • I do agree with your “averaging machine” argument. It makes a lot of sense given how LLMs are trained as essentially massive statistical models.

    For image generation models I think a good analogy is to say it’s not drawing, but rather sculpting - it starts with a big block of white noise and then takes away all the parts that don’t look like the prompt. Iterate a few times until the result is mostly stable (that is it can’t make the input look much more like the prompt than it already does). It’s why you can get radically different images from the same prompt - the starting block of white noise is different, so which parts of that noise look most prompt-like and so get emphasized are going to be different.


  • People shit on Hossenfelder but she has a point. Academia partially brought this on themselves.

    Somehow I briefly got her and Pluckrose reversed in my mind, and was still kinda nodding along.

    If you don’t know who I mean, Pluckrose and two others produced a bunch of hoax papers (likening themselves to the Sokal affair) of which 4 were published and 3 were accepted but hadn’t been published, 4 were told to revise and resubmit and one was under review at the point they were revealed. 9 were rejected, a bit less than half the total (which included both the papers on autoethnography). The idea was to float papers that were either absurd or kinda horrible like a study supporting reducing homophobia and transphobia in straight cis men by pegging them (was published in Sexuality & Culture) or one that was just a rewrite of a section of Mein Kampf as a feminist text (was accepted by Affilia but not yet published when the hoax was revealed).

    My personal favorite of the accepted papers was “When the Joke Is on You: A Feminist Perspective on How Positionality Influences Satire” just because of how ballsy it is to spell out what you are doing so obviously in the title. It was accepted by Hypatia but hadn’t been published yet when the hoax was revealed.



  • … because they didn’t.

    …and when they do, it becomes because they don’t have a large enough majority. If they had a filibuster -peoof majority, it would be because they don’t have a veto-proof majority even if they also held the executive.

    To put it another way, Biden had the power to fully legalize marijuana at any time because drug scheduled can be changed by process from either the legislative or executive branches. Instead he had it rescheduled. Because Dems only operate in half measures federally, because they are controlled opposition.