• 1 Post
  • 70 Comments
Joined 2 years ago
cake
Cake day: April 27th, 2024

help-circle


  • Actually, I agree. And so far, small local models are really solid, and can punch above its weight even when compared to frontier models.

    I believe what I meant when I said I doubted it was since these AI corpos seemingly give no indication that local is an option, so most people would think they can only access an LLM through the web. This would bolster the SaaS ecosystem dominating over local AI, although local will keep increasingly growing as a more favourable option.

    Although I do agree that the industry will shift from being server based to PC based inference as well, I don’t see that shift being large enough to make these companies change their training paradigms to include telemetry from local AI, but I’m sure some will.



  • I doubt it, but honesty, many systems can do inference pretty well, like how I ran the MLX version of Qwen 3 4b with a DuckDuckGo search RAG, and used it to ask quick questions and verify some simple things, running on a MacBook Air m2 16gb, and barely made a dent in the RAM utilisation or SoC, and this also goes for my much less powerful machines, like even a galaxy a20, with 3gb of memory and a low spec octacore exynos, can run small models really well, although the quantisation needs to be a bit strict.






  • When I was in public high school last year, the boys toilets were horrid, someone spilled baby oil all over the place for some reason, people would rip off doors, smash the toilets, make wall smear murals, leave their CO2 canisters everywhere (which was what they’d get high off of, odd choice but an interesting one I suppose), and of course, piss literally anywhere. I mean, hey, at least the massive mirror was completely intact somehow…


  • I agree on that front, the engineering of these bins has gotta be a big factor in the cost, although there has been government and union corruption recently, which would’ve ballooned the cost a little, but I’m not sure by how much.

    what I don’t understand however is why they chose to do a machete amnesty program rather than anything else, such as improving rehabilitative justice, and removing the endless crime loophole caused by bailouts (which tbf is being acted on currently,) for example.

    Edit: now that I think of it, its probably since the government might want free credit for “solving the crime epidemic”, even though they’re mostly half solutions




  • We should all explore more, not stay stuck in a bubble of 6 or so megacorp websites that take most of the worlds traffic.

    The internet is distributed, with billions of websites that are all unique.

    Even if it’s difficult to find new websites to look at, then choose a different internet medium such as dark net browsers such as tor and i2p to explore the more privacy respecting web, or on the lighter side, go for geminispace or gopher browsers like Lagrange to explore the un-corporatised early internet style web.

    It’s hard to explore when you don’t know where to look.


  • The algorithms in these social media services don’t care about you actually wanting to watch the video or see the content in general, instead they hyperfixate on if you took 0.02 seconds longer before scrolling to the next one compared to the previous scrolls, to determine what should be shown.

    For example, if your interests are exclusively in random gadgets and trinkets for example, but then it shows you a video of an onlyfans promotion, you may accidentally pause in confusion, then scroll, unknowingly triggering the algorithm to keep showing you onlyfans promotions despite you not wanting that.

    And it’s the same if it’s offensive or triggering, the algorithm decides to keep showing you such content so it can stun you into staying in the app longer.

    This is why I’ve decided to outsource my social media use to lemmy and geminispace BBS boards, because sure instagram can be enjoyable a lot of the time, but if you use it excessively, it’s damaging.

    Also, pro tip: if you don’t use Insta for a month or so, it decides to show you the best content possible to try getting you hooked again, but once you get to a video which isn’t awfully enjoyable, close it and forget about it for another month. Mileage may vary but it works for me somehow, even if I’m likely to be AuDHD