Interesting read, thanks! I’ll finish it later, but already this bit is quite interesting:
Without access to gender, the ML algorithm over-predicts women to default compared to their true default rate, while the rate for men is accurate. Adding gender to the ML algorithm corrects for this and the gap in prediction accuracy for men and women who default diminishes.
We find that the MTEs are biased, signif-icantly favoring White-associated names in 85.1% of casesand female-associated names in only 11.1% of case
If you’re planning to use LLMs for anything along these lines, you should filter out irrelevant details like names before any evaluation step. Honestly, humans should do the same, but it’s impractical. This is, ironically, something LLMs are very well suited for.
Of course, that doesn’t mean off-the-shelf tools are actually doing that, and there are other potential issues as well, such as biases around cities, schools, or any non-personal info on a resume that might correlate with race/gender/etc.
I think there’s great potential for LLMs to reduce bias compared to humans, but half-assed implementations are currently the norm, so be careful.
After all these years, I’m still a little confused about what Forbes is. It used to be a legitimate, even respected magazine. Now it’s a blog site full of self-important randos who escaped from their cages on LinkedIn.
There’s some sort of approval process, but it seems like its primary purpose is to inflate egos.
It was an SEO hellhole from the start, so this isn’t surprising.
Do Forbes next!
It says:
Available Architectures
aarch64, x86_64
And it uses Android Translation Layer. Interesting. I’ll give it a shot on my desktop later.
Yeah, they were able (and thus legally required) to hand over the user’s recovery email address, which is what got them caught. You don’t need to enter a recovery email address, and you can of course choose to use an equally-secure service for recovery.
One big technical issue to note is that Proton doesn’t use end-to-end encryption for email headers, which includes recipients and subject lines, among other things. So that’s potentially exposed to law enforcement as well. I believe Tuta does encrypt headers.
I don’t think the free accounts cost them a whole lot. You only get 1GB of mail space, and the free versions have minimal features (e.g. you can only create one email filter). They make their money on paid accounts, which seems legit to me.
simply logging out or using an alt account
It is increasingly difficult to use X without an account. Not sure what the signup process is like nowadays. IIRC it used to require phone number verification in the Twitter days, but perhaps Musk relaxed the requirements in order to better pad the usage stats with spambots?
Yeah, AMD is lagging behind Nvidia in machine learning performance by like a full generation, maybe more. Similar with raytracing.
If you want absolute top-tier performance, then the RTX 4090 is the best consumer card out there, period. Considering the price and power consumption, this is not surprising. It’s hardly fair to compare AMD’s top-end to Nvidia’s top-end when Nvidia’s is over twice the price in the real world.
If your budget for a GPU is <$1600, the 7900 XTX is probably your best bet if you don’t absolutely need CUDA. Any performance advantage Nvidia has goes right out the window if you can’t fit your whole model in VRAM. I’d take a 24GB AMD card over a 16GB Nvidia card any day.
You could also look at an RTX 3090 (which also has 24GB), but then you’d take a big hit to gaming/raster performance and it’d still probably cost you more than a 7900XTX. Not really sure how a 3090 compares to a 7900XTX in Blender. Anyway, that’s probably a more fair comparison if you care about VRAM and price.
Basically the only thing that matters for LLM hosting is VRAM capacity
I’ll also add that some frameworks and backends still require CUDA. This is improving but before you go and buy an AMD card, make sure the things you want to run will actually run on it.
For example, bitsandbytes support for non-CUDA backends is still in alpha stage. https://huggingface.co/docs/bitsandbytes/main/en/installation#multi-backend
You can root Graphene if you want to, right?
IIRC, Android has always had native support for keyboards and mice. I remember connecting a bluetooth mouse to my old Nexus 4 running…Android 4, maybe 5? It worked out of the box. Saved my butt when the touch screen broke. :)
Can’t say I’ve tried this in recent years but I think it still works, yeah?
I’ve never tried it myself, but I think you can run full Linux VMs on Pixel phones already. A quick search brings up https://www.xda-developers.com/nestbox-hands-on/
Anyone have experience with this or similar options? Personally I’ve never used anything more advanced than Termux (which is lean and super cool, but not a full-blown VM).
Yep. If it uses a cloud service, they’re probably going to squeeze you, pull a bait-and-switch, or go out of business. The only exceptions that spring to mind are services with significant monetization in the corporate space, like Dropbox. And I’m not really confident that Dropbox’s free tier will remain viable for long, either.
Even non-cloud-based apps are risky nowadays because apps don’t remain compatible with mobile OSes for very long. They require more frequent updates than freeware/shareware generally did back in the 90s. I remember some freeware apps that I used for 10 years straight, across several major OS versions, starting in the 90s. That just doesn’t happen anymore. I’ve been using Android for over 10 years and I don’t think there’s a single app I used back then that would still work.
Single-purchase apps are basically dead, at least on mobile platforms. Closed-source freeware is dead, too. If it’s open-source, if push comes to shove someone can always pick up the torch and update it. It’s very rare for an open-source project to be completely abandoned without there at least being a viable open-source alternative available.
At this point, I don’t even look at Google Play. It’s F-Droid or bust.
Hmm, I’ll have to double-check it on my build. Pretty sure it was working, but it’s possible I was only getting OpenCL or Vulkan acceleration. Out of curiosity, what’s your GPU model? I think only the 7900XT[X] is officially supported for ROCm.
I recently tried Bazzite, and I have to agree. Switching from a traditional Linux distro to an immutable distro is harder than switching from Windows to Linux. I’m not kidding. When it comes to immutability, my experience can be split into two general cases:
I have yet to encounter a scenario where immutability offered a tangible benefit. The supposed advantages seem rather abstract. I can’t break my system? Okay…but…well, I already had snapper for the rare occasions when something got royally borked. This is a problem that has already been solved without major compromises, so why are we now compromising so much to solve it again?
It comes with 4 different package management systems (or 6 if you count Distrobox and Waydroid), and they all come with big caveats. I’ve had to reboot more in the past week than I previously had in the past year on Debian, because every time I need to install something from the main Fedora repo with rpm-ostree (which has been many times already), it needs to reboot. They recommend against using rpm-ostree, but there is no reasonable alternative for a rather wide array of software. It’s either rpm-ostree or build a whole mess of things from source and manage them manually. Both options suck very hard.
Still, overall, Bazzite delivers. Everything you see on their web site works out of the box. It’s hard to recommend, but it’s also hard to criticize. I’ve never had a smoother gaming experience, and this is the first time I’ve ever had to spend zero minutes configuring my GPU drivers (outside of macOS, anyway). You get CUDA and ROCm out of the box. You get the latest drivers. It’s awesome.
If you’re wondering if an immutable distro is right for you, the answer is probably “no”. But if you’re up for the, erm, “adventure” of learning this new paradigm, Bazzite fucking rocks.
I’m certain that if someone did collect data from the Fediverse; it would become a hot topic
I’d assume bad actors (or at least chaotic neutral actors) are slurping up the entire fediverse already. It is trivial to do, and nobody would know.
I mean, the whole point is that anyone can spin up a server and federate with others. I could start my own server, which would by default federate with almost all other servers. That means I wouldn’t even need to write a scraper. All that data would be sent straight to my server. All I need is access to my own database at that point. With Lemmy, I’d even get users’ upvote/downvote history, which is not visible in any clients AFAIK. The only barrier would be to subscribe to communities on different servers to kickstart federation.
As long as you don’t run obvious spam/bot accounts, nobody would block your instance.
Alternatively, if you want to write a scraper, that’s also pretty easy. Most servers are publicly accessible. Every community has an RSS feed. You don’t even need an account in general. Again, the whole point is to be open and accessible, in contrast to closed-off data-misers like Facebook, Reddit, and X.
The fediverse is friendly to users, with very little regard for what those users might do. I believe this is the correct philosophy, but I won’t pretend that it doesn’t leave us open to bad behavior.
So probably there will be some systems other than Linux that do use Rust
There’s one called Redox that is entirely written in Rust. Still in fairly early stages, though. https://www.redox-os.org/
What part of this is misinformation, exactly? Seems pretty well-supported.
Wait, isn’t it the other way around? You should arrive in NY earlier than you left London, since NY is 5 hours behind London. So if you leave at 8:30 and arrive 1.5 hours later, it should only be 5AM when you arrive.
You might need a third breakfast before your elevenses in that case.