Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)M
Posts
1
Comments
142
Joined
2 yr. ago

  • Part of the problem here is that AI is mostly done by companies with billions of investments and in turn they NEEEEEDDDDD engagement, so they all made their AI as agreeable as possible just so people would like it and stay, with results like these becoming much more "normal" than it should or could be

    I wonder how much of that is intentional vs a byproduct of their training pipeline. I didn't keep up with everything (and those companies became more and more secretive as time went on), but iirc for GPT 3.5 and 4 they used human judges to judge responses. Then they trained a judge model that learns to sort a list of possible answers to a question the same way the human judges would.

    If that model learned that agreeing answers were on average more highly rated by the human judges, then that would be reflected in its orderings. This then makes the LLM more and more likely to go along with whatever the user throws at it as this training/fine-tuning goes on. Instead of the judges liking agreeing answers more on average, it could even be a training set balance issue, where there simply were more agreeing than disagreeing possible answers. A dataset imbalanced that way has a good chance of introducing a bias towards agreeing answers into the judge model. The judge model would then pass that bias onto the GPT model it is used for to train.

    Pure speculation time: since ChatGPT often produces two answers and asks the user which one the user prefers, I can only assume that the user in that case is taking the mantle of those human judges. It's unsurprising that the average GenAI user prefers to be agreed with. So that's also a very plausible source for that bias.

  • I don't know too much about Minecraft and especially not Sky Block. But isn't the implication also that they wasted their Lava by turning it into an Obsidian instead of using water+lava to create a cobble stone generator, thus softlocking their progress entirely?

  • Coincidentally enough, "noon" is etymologically related to "nine", so 10 is indeed afternoon in a very literal sense.

    (Just ignore the fact that it's originally meant to be "the ninth hour after sunrise", so ca. 3pm)

  • Did they? All I remember him doing, and all I can find him doing, is:

    • Swear on Arthur's name
    • Mention a Loxley who is quick/fast
    • Use Excalibur
    • One tiny hint that might actually give him away entirely, that I cannot point out specifically

    In the credits he's also listed as "Seiba".

  • why is it dumb?

    The dumb bit is them revealing his true name right away while the related anime's pre-release episode, which they're intentionally syncing up with, is putting in some effort to hide it. I'd wager plenty of people have not read FsF, although maybe I'm mistaken there. Additionally, FGO has shown that they already have the capability of not revealing true names until the time is right before, so it's not like this would've needed some crazy new tech.

    As for your last paragraph, I was talking about this FsF Saber, not OG Saber.

  • We have quite a few more Sabers that use red as a primary or secondary color in FGO. That being said, I'd recommend not looking into that because for some dumb reason they released this Saber a year ago for New Year's with his full name revealed and everything (actually coinciding with the preview release of this very episode).

    Having read the FsF LN (as far as it is released, one volume missing still), I think Saber is one of the few servants in FsF where giving away their true name too early actually detracts from the experience a little.

  • The nomination phase has a few suggestions for you, based on what you played. But if you don't like them/want something else, there's a button for that. Now you'll know for next year.

  • Deleted

    Permanently Deleted

    Jump
  • I was taught to spell out until twelve, after that the *teen ones start, which are a lot more letters and space on the paper.

    Of course it all depends on whatever the relevant style guide says. Consistency is the most important bit.

  • It's two jokes. Firstly, "Barbarian" was a ancient Greek-"invented" term for people who don't speak Greek (or heavy dialects of Greek). The generally accepted theory is that "Barbar" is them imitating sounds they don't understand, similar to a modern "blabla". Secondly, Hercules is the Roman name for the Greek Heracles.

  • My uni had one. Sadly I couldn't fit it into my schedule because of overlaps and other requirements.

  • In the OOP's example, that is solved by magecraft (which is distinct from "true" magic btw) losing potency by becoming general knowledge, thus forcing mages into working their magecraft in secrecy.

    Both the mages and the Church work hard at keeping it a secret, albeit with different motives and methods.

  • *With AI review :)

  • While it doesn't say anything about IIV specifically, they sure got creative enough to sometimes subtract more than one of the smaller units from a larger one.

  • I kept up with the drama until about a week ago so what I'm saying here is the status from back then. Someone please add any new context if I'm missing any new developments:

    From what it appeared, view counts dropped but ad revenue stayed the same. Even before this whole thing, YouTube pays out for ads watched (and clicked). Pay out was not dependent on raw view count for a long time, if ever.

    This suspicious behavior of view count dropping but ad revenue staying the same is actually what tipped people off that the issue was adblock related. The fact that channels with a larger focus on a younger audience seeing less of a drop also helped.

    Now those view counts dropping could still have an indirect, negative effect on ad revenue, if it, e.g. automatically leads to YouTube recommending their videos less prominently.

  • This is also sorta how RAW works (in DnD 5e), to quote the PHB:

    Group ChecksWhen a number of individuals are trying to accomplish something as a group, the DM might ask for a group ability check. In such a situation, the characters who are skilled at a particular task help cover those who aren't.To make a group ability check, everyone in the group makes the ability check. If at least half the group succeeds, the whole group succeeds. Otherwise, the group fails.

    Taking the median roughly has the same effect, it only has a chance to differ if the number of successes and the number of failures are tied.

  • Another European here to chime in that l also learned to write capital As like that in cursive.

    The rs, fs and ts don't look like how we were taught though.

  • I usually just go with 1.5 because adding half/subtracting a third is way easier to do in my head, and I'm not worried about a ~10% error in casual conversation.

  • I've been to multiple museums in Japan (which is somewhat relevant because Nintendo is Japanese) that either flat out ban all photography (e.g. Ghibli Museum, Aomori Museum of Modern Art) or have some exhibits that you're not allowed to take pictures of (e.g. Tokyo National Museum). One exhibit I wanted to take a picture of had a "no photography" sticker on it, but it was on the opposite side from where I approached so I didn't see it, causing staff to run up to me when I pulled out my phone to point out the sign.

    I've also heard from other tourists that "no photos" seems to be rather common there.

    Btw, I'm not at all saying that they're justified at all, just saying that there are indeed places that forbid photos for copyright reasons. In my opinion, no photo would ever match seeing the exhibits in person so it is entirely pointless to ban them. Even professional, official scans of pieces don't come close.

  • Me and all my friends all call them "Handys".

    And to people not familar with English loanwords in German: Yes, the correct German pluralization ends with a "ys" and not as it would be correct in English: "ies". The same is true for "Hobbys" and "Babys", not sure if there are more.

  • 196 @lemmy.blahaj.zone

    r(UwU)le