• 0 Posts
  • 9 Comments
Joined 11 months ago
cake
Cake day: August 2nd, 2023

help-circle
  • Any chance you have an nvidia card? Nvidia for a long time has been in a worse spot on Linux than AMD, which interestingly is the inverse of Windows. A lot of AMD users complain of driver issues on Windows and swap to Nvidia as a result, and the exact opposite happens on Linux.

    Nvidia is getting much better on Linux though, and Wayland+explicit sync is coming down the pipeline. With NVK in a couple years it’s quite possible that nvidia/amd Linux experience will be very similar.


  • Nevoic@lemm.eetoTechnology@lemmy.worldHello GPT-4o
    link
    fedilink
    English
    arrow-up
    8
    arrow-down
    3
    ·
    2 months ago

    “they can’t learn anything” is too reductive. Try feeding GPT4 a language specification for a language that didn’t exist at the time of its training, and then tell it to program in that language given a library that you give it.

    It won’t do well, but neither would a junior developer in raw vim/nano without compiler/linter feedback. It will roughly construct something that looks like that new language you fed it that it wasn’t trained on. This is something that in theory LLMs can do well, so GPT5/6/etc. will do better, perhaps as well as any professional human programmer.

    Their context windows have increased many times over. We’re no longer operating in the 4/8k range, but instead 128k->1024k range. That’s enough context to, from the perspective of an observer, learn an entirely new language, framework, and then write something almost usable in it. And 2024 isn’t the end for context window size.

    With the right tools (e.g input compiler errors and have the LLM reflect on how to fix said compiler errors), you’d get even more reliability, with just modern day LLMs. Get something more reliable, and effectively it’ll do what we can do by learning.

    So much work in programming isn’t novel. You’re not making something really new, but instead piecing together work other people did. Even when you make an entirely new library, it’s using a language someone else wrote, libraries other people wrote, in an editor someone else wrote, on an O.S someone else wrote. We’re all standing on the shoulders of giants.


  • Nevoic@lemm.eetoTechnology@lemmy.worldHello GPT-4o
    link
    fedilink
    English
    arrow-up
    23
    arrow-down
    5
    ·
    edit-2
    2 months ago

    18 months ago, chatgpt didn’t exist. GPT3.5 wasn’t publicly available.

    At that same point 18 months ago, iPhone 14 was available. Now we have the iPhone 15.

    People are used to LLMs/AI developing much faster, but you really have to keep in perspective how different this tech was 18 months ago. Comparing LLM and smartphone plateaus is just silly at the moment.

    Yes they’ve been refining the GPT4 model for about a year now, but we’ve also got major competitors in the space that didn’t exist 12 months ago. We got multimodality that didn’t exist 12 months ago. Sora is mind bogglingly realistic; didn’t exist 12 months ago.

    GPT5 is just a few months away. If 4->5 is anything like 3->4, my career as a programmer will be over in the next 5 years. GPT4 already consistently outperforms college students that I help, and can often match junior developers in terms of reliability (though with far more confidence, which is problematic obviously). I don’t think people realize how big of a deal that is.





  • Nevoic@lemm.eetoTechnology@lemmy.worldTesla scraps its plan for a $25,000 Model 2 EV
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    6
    ·
    edit-2
    3 months ago

    Depends on what you’re looking for. I had a high paying tech job (layoffs op), and I wanted a fun car that accelerates fast but also is a good daily driver. I was in the ~60k price range, so I was looking at things like the Corvette Stingray, but there are too many compromises for that car in terms of daily driving.

    The Model 3 accelerates faster 0-30, and the same speed 0-60. Off the line it feels way snappier and responsive because it’s electric, and the battery makes its center of gravity lower, so it’s remarkably good at cornering for a sedan, being more comparable to a sports car in terms of cornering capabilities than a sedan.

    Those aren’t normally considerations for people trying to find a good value commuter car, so you would literally just ignore all those advantages. Yet people don’t criticize Corvette owners for not choosing a Hyundai lol

    On the daily driving front, Tesla wins out massively over other high performance cars in that price range. Being able to charge up at home, never going to a gas station, best in class driving automation/assistance software, simple interior with good control panel software, one pedal driving with regen breaking.

    If you’re in the 40k price range for a daily commuter, your criteria will be totally different, and I am not well versed enough in the normal considerations of that price tier and category to speak confidently to what’s the best value. Tesla does however, at the very least, have a niche in the high performance sedan market.


  • Nevoic@lemm.eetoTechnology@lemmy.worldTesla scraps its plan for a $25,000 Model 2 EV
    link
    fedilink
    English
    arrow-up
    9
    arrow-down
    9
    ·
    edit-2
    3 months ago

    Like sure fuck Elon, but why do you think FSD is unsafe? They publish the accident rate, it’s lower than the national average.

    There are times where it will fuck up, I’ve experienced this. However there are times where it sees something I physically can’t because of either blindspots or pillars in the car.

    Having the car drive and you intervene is statistically safer than the national average. You could argue the inverse is better (you drive and the car intervenes), but I’d argue that system would be far worse, as you’d be relinquishing final say to the computer and we don’t have a legal system setup for that, regardless of how good the software is (e.g you’re still responsible as the driver).

    You can call it a marketing term, but in reality it can and does successfully drive point to point with no interventions normally. The places it does fuckup are consistent fuckups (e.g bad road markings that convey the wrong thing, and you only know because you’ve been on that road thousands of times). It’s not human, but it’s far more consistent than a human, in both the ways it succeeds and fails. If you learn these patterns you can spend more time paying attention to what other drivers are doing and novel things that might be dangerous (people, animals, etc ) and less time on trivial things like mechanically staying inside of two lines or adjusting your speed. Looking in your blindspot or to the side isn’t nearly as dangerous for example, so you can get more information.


  • Nevoic@lemm.eetoLinux@lemmy.mlLinux hits 4% on the desktop 🐧 📈
    link
    fedilink
    arrow-up
    8
    arrow-down
    2
    ·
    edit-2
    4 months ago

    Linux is a far more reliable operating system at the kernel level, which is why the vast majority of the Internet runs on Linux, and is very stable compared to anyone’s personal computer (no matter O.S). It’s also lighter weight at its core, which is a big plus for servers.

    The thing about Linux desktops that tend to be finicky is interop with some proprietary software (e.g nvidia drivers) or desktop environments (gnome can freeze/crash if you like running bleeding edge before bugs are ironed out). Windows has issues too however, free software often literally doesn’t run on Windows (requiring WSL, the same way games on Linux require wine), and the desktop environment is essentially indistinguishable from the base operating system. When you get a desktop environment crash on Windows, your system will BSOD and restart with no recourse, in Linux I can ssh into my still functioning computer and kill my DE, or drop to the TTL and do the same thing.

    The end might not seem like a big deal for some people (who cares if you have to restart by a button press or kill your DE and login, they’ll take a similar amount of time), but for someone like me where reliability is a big concern (as in, uptime for the half a dozen services/containers I run for people), this is great. People watching media off of jellyfin don’t have to stop because of a DE bug, but on Windows a BSOD would stop their media (and within the last week we’ve had several BSODs on Windows PCs due to bugs relating things like adaptive sync or sometimes just unknown reasons).

    For what it’s worth I also game exclusively on Linux, vk3d, dxvk, and proton are godsends. Somethings don’t work, developers who won’t flip the switch for EAC (e.g Fortnite), but for me the games I play always worked. This will actually change soon, Vanguard is coming to League and that only works on Windows, but also probably not my last install of Windows (I tried W11 when it came out because I’m just curious about new tech), but I had to do a TPMBypassCheck despite having ftpm enabled in the BIOS, and afaict, at least from people I know with similar builds to me, if this happened then firmware TPM probably isn’t being picked up by W11, and that means I need to buy a TPM module or drop to W10 to play League. Plus, vanguard is an intense rootkit with full 24/7 access to your O.S so I probably don’t want that installed anyway, even if it happened to work on Linux. Just going to stick to SoD for now in my free time lol