• 0 Posts
  • 42 Comments
Joined 1 year ago
cake
Cake day: July 6th, 2023

help-circle
  • I had to do a lot for my dog that could be considered gross in an abstract sense but it never bothered me much. I suppose that was because he was my cute dog and not just because he was a cute dog, but being a cute dog certainly helped. It’s notable that I did not feel that same way about my little sister when I was 11 and she was born. She’s a great person now but I didn’t like helping take care of her even though I didn’t have to do it much; I think that’s a big reason why I don’t have my own children.



  • ArbitraryValue@sh.itjust.workstoLemmy Shitpost@lemmy.worldJust don't do it
    link
    fedilink
    English
    arrow-up
    8
    arrow-down
    2
    ·
    edit-2
    5 hours ago

    A friend of mine has a baby and he’s certainly in that joy category. I guess it shouldn’t surprise me because it was essential for the survival of the species, but I’m still surprised that people like babies. Kids can be cute when they’re old enough to talk and use the toilet, and even then they’re only cute if they’re the charming kids-say-the-darnest-things sort and not the tantrum-throwing sort. Meanwhile puppies and kittens are cute from the moment they’re born.

    (Why do humans babies make such horrible sounds when they need something? A puppy whimpering makes me want to give it whatever it’s asking for. A baby crying makes me angry. I suppose the crying is harder to simply ignore.)


  • I don’t think that’s an impossible problem. Existing models can reliability distinguish between, for example, different languages. Most of their training data is presumably in English but while this may make them better at generating English text, it doesn’t make them randomly switch from other languages to English. A sufficiently advanced model would likewise distinguish between descriptions of reality and shit-posts because the content of shit-posts would not be useful for predicting descriptions of reality. Some fine tuning would teach it to produce just the descriptions of reality.

    Or look at it this way: the folks developing these LLMs aren’t ignorant of the fact that Reddit content is often false and meant to be funny. They’re not going to make the sort of silly mistake that someone who isn’t an expert can still easily predict and they’re not going to train their LLMs on that content if it makes the LLMs worse, although we’re still going to see some glue on pizza while the technology continues to develop.




  • defense attorneys argued that Manhattan prosecutors had placed “highly prejudicial emphasis on official-acts evidence,” including Trump’s social media posts and witness testimony about Oval Office meetings

    It’s unclear to me why an official act cannot be used as evidence that a different unofficial act occurred. Let’s say candidate Trump shoots Bob on Fifth Avenue and then, after being elected, threatens to “kill Joe the way [he] killed Bob” during his State of the Union address. He can’t be held accountable for threatening to kill Joe, but he did just confess that he killed Bob while he wasn’t president. Why couldn’t this confession be used as evidence in his trial for killing Bob? Or, for that matter, in his trial for killing Joe if he went on to kill Joe after he was out of office?






  • When I was in elementary school, my parents would get me abridged versions of great books from Walmart. They were little paperbacks with lots of illustrations. I loved many of them and read them over and over. Then when I got older, I read many of the originals.

    I think that some good stories can be retold in many different ways. One telling will be better than another but even an abridged telling can preserve they key pieces and convey them to a different audience.

    Edit: consider folk-tales. They don’t have a canonical version and so for example we can have Robin Hood in both Water Scott’s Ivanhoe and in the Disney movie with the foxes. Or Greek mythology, which can be enjoyed even if you’re not reading Hesiod.






  • ArbitraryValue@sh.itjust.workstopolitics @lemmy.worldCNN's debate was no fair fight
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    1
    ·
    edit-2
    6 days ago

    Then it becomes “okay, call this prick the c-word. Now I need to also cite this fact that is part of my border security answer. And then I need to talk about… jesus christ are we actually talking about global warming right now?”

    That would be an understandable reaction from the average person but the president should be a lot more capable than the average person. Even if this specific sort of thing isn’t something he needs to be able to handle, he still needs to handle things a lot harder than this and his performance here isn’t reassuring me that he can. Trump is so predictably rude that Biden should have been totally ready for it.


  • I should clarify. I’m not saying that most people who distrust the justice system are going to like Trump more after his conviction. I’m also not saying that I think he’s likely to reform the justice system in a way that helps people affected by racial bias.

    However, many of Trump’s supporters consider his conviction evidence that he’s genuinely an anti-establishment candidate rather than proof of wrong-doing. (See the variety of “I’m voting for the convicted felon” merchandise.) This attitude requires a distrust of the justice system. We’ve already seen that Trump’s conviction hasn’t hurt his poll numbers very much and that he currently has more black support than he did in '16 or '20 so I’m saying that his conviction might actually lead to a small increase in support for him from black people (the majority of whom are still never going to support him) because more of them distrust the justice system.