• 2 Posts
  • 1.61K Comments
Joined 3 years ago
cake
Cake day: June 14th, 2023

help-circle
  • While it may not compare to the closed, corporate “frontier” models that are running each query on racks upon racks of servers worth millions of dollars in a corporate data center, it’s not clear to me that’s a reasonable comparison. Mistral is actually one of my personal favorite models and it has nothing to do with being French (although that is a huge bonus, it doesn’t really contribute to my method of scoring and ranking of different models). It’s simply more reliable and consistent than a lot of the other open-weight models, it follows directions well and has good task comprehension. Even with abliteration (guardrail removal) it doesn’t generally have weird artifacts and it uses diverse, clear and reasonable language in its responses. It also handles Markdown really nicely in a very consistent way without needing guidance. To me, these are more important properties than simply “containing the most recent training data” or how many agents it has. At the end of the day I’m going to judge a model based on its output more than how it gets there and it turns out you don’t need all the latest whiz-bang features to get acceptable output depending on what exactly you’re doing with it.

    It may not be as technically impressive as even the latest greatest open-weight models like Qwen, but it’s still a great model whether it’s French or not, and I would still use it just as often as I already do even if it were Chinese or North Korean or Russian I don’t care, all that matters is that it’s open-weight, runs with reasonable hardware requirements, is flexible in how it can be used, and produces good output. Mistral ticks all the boxes.







  • I can see the appeal of eschewing society as a whole in favor of a bot that I know won’t ask me my political or religious leanings and then tailor their behavior around my answer.

    … tailoring their behavior around your answer is literally how LLMs work. That’s why they’re so sycophantic. Also unless you’re running it locally on a machine you control, it doesn’t need to ask you about your political or religious leanings anyway because it already knows. That’s exactly the sort of context that data brokers have already long-since developed around your identity, and a commercial AI model is absolutely going to be looking at that kind of context to know exactly how to talk to you.

    You are going the wrong direction if you think AI is the solution to any of these things at least in the way it is currently being used.

    Casual smalltalk with randos is probably the cure. “Much less appealing” is the environment that’s been intentionally created to prevent you from doing that. We’re all in the same boat.


  • Tilley Endurables is a Canadian company that was owned by a company in the UK for awhile and is now back in Canada. They are famous for their hats, and they used to have a no-questions-asked, pretty much unqualified lifetime replacement guarantee on (some of) their clothes and actually challenged their customers to find ways to try to wreck them in actual use, with the washing instructions famously being “give them hell”. But that’s all been enshittified to the point that most of their stuff isn’t guaranteed anymore and even the stuff that is, is effectively only guaranteed for meaningless “normal wear and tear and defects” nowadays. In my experience they’re still pretty decent quality and very rugged, or at least they were when I last bought them. Although I don’t have a ton of confidence that they’ve stayed that way or will continue to, they still seem to be from what I can tell.








  • cecilkorik@lemmy.catoSelfhosted@lemmy.worldKitchenOwl Gone?
    link
    fedilink
    English
    arrow-up
    16
    arrow-down
    8
    ·
    3 days ago

    He’s still right that it’s weird people like you are going to bat to defend them. Microsoft sucks. It must get tiring if you have to call out every inaccurate thing everyone says to try to tear them down. The important take-home message is that we need to tear them down, they suck.

    You don’t see people bothering to defend Epstein, for example. Even though there’s lots of inaccurate stuff going around, there’s enough accurate stuff to be absolutely confident he was an absolute loathsome piece of shit not worth defending. Not worth the effort to defend. Why bother?

    What do you see in Microsoft that you think is worth defending? Github is shit, and it’s evil. Let it go.



  • We have to stop accepting this as a fact of life. It’s only going to change if we force it to. The more people who start asking questions like this and being politically active in their lives, the more things actually will eventually begin to change. Pursue reason and rationality, demand it from our government. It isn’t fun and it feels impossible so we treat it like a joke. Our mockery and nihilism is comfortable but ineffective, our complacency serves no one. It isn’t impossible, we’ve just decided it is because we’ve never really tried to fight for it and we are constantly reminded of the times we failed, any times we succeed are quickly normalized while any influence the people might’ve had on the outcome is minimized and dismissed. We are much more powerful than we are allowed to know.