I think a more reasonable assumption (than trying to destroy the fediverse) would be marketing or propaganda - somebody creates users using LLMs, gets them accepted, legitimizes them by generating interactions, then either sells the accounts or uses them to sell a service to have them promote something, with a history that looks like a real user.
The screenshot looks like it might be a cloudflare verification page, which would put another layer of separation between the site owner and the QR system.
People looking up alternatives doesn’t mean they actually want to use them. Switching to an alternative necessarily means some pains, in terms of changed interface, functionality, and community. You can’t just magically switch between platforms and have all the people follow you there, and that’s a dealbreaker for many.
As far as I know, the fediverse is growing, just not explosively. We can hope that at some point it reaches a threshold that convinces a lot of people to move over, cementing it as a “proper” platform.
That’s presumably (and hopefully) not GenAI, but a much lighter classification model built with the sole purpose of judging if an image is problematic, I have no problem with those.
*sweaty
Opinionated sweater is when somebody offers to refund the sweater they gave you as a gift
Reminds me of a recent xkcd: https://xkcd.com/3201
One point on perception - doesn’t the sun appear somewhat yellow because the blue light has a stronger tendency to scatter, meaning that the roughly white light of the sun is less blue, with all the blue color of the sky being taken away from the color of the sun?
I’ve gotten such symptoms before when running out of RAM - I’m on Arch and never bothered setting anything up for that instance and I’m not sure what’s going on, but I think the system is struggling to recover memory or something before it resorts to killing processes, and would sometimes freeze for a minute like that.
That said, yeah… Kernel modules (which device drivers often are) are allowed to run at a higher level of privilege, with less oversight, more access to hardware and better performance, so if they misuse that privilege they can break things badly. And with proprietary drivers, you have no idea or control of what it’s actually doing, so you can only try to downgrade or wait and hope the company fixes it.
Hell, I’ll deadass reheat my food mid-meal if it starts going cold, so I can enjoy the last bits just as much.
The weakest part of any security system is the people.
Well, maybe not any, but most ;D
I’ve got one light in a room that makes a quiet whining noise when on, seemingly only after a minute or so (maybe after it warms up a bit). Thankfully I can just keep it off just fine, but occasionally I’ll turn it on for a bit more brightness, and realise it’s still on a while later by the annoying noise.
I don’t think “update notes or any marketing material” qualifies for making this kind of change non-silent - if the update is pushed through the same channel as regular security updates, and doesn’t explicitly notify the user the behavior of the button has changed, that’s pretty silent.
Often for those kinds of updates software will show a special introduction screen, tutorial, or outright a prompt asking you to choose between the new and old behavior - but that’s software from people that care about the user having a good experience, and making such changes is a big deal for them.
You can call “bs” all you want, but you’re yelling at the messenger. If I understand the situation correctly, you’re using an Amazon service, so Amazon promoting Amazon stuff on that service might count as self-promotion (or whatever the term is) rather than advertisement. The difference being that they’re not being paid by somebody to promote their thing, and just calling out their own services/events.
Note that I’m also not supporting this, I don’t like it, but I’m also not surprised if that’s how that works, and that Amazon would be using it this way.
If I understand correctly, it’s a different kind of “immutable”, since distros like Bazzite provide premade immutable images you use and anything else you need you install using alternative means, whereas NixOS is an immutable image generator that requires you to set up your own definitions for the image, but also lets you install software by adding it to that image.
In-memory kernel patching is complicated, AFAIK only select distributions support it, right? If kernel hotswap is successfully implemented this way, it should allow switching between arbitrary kernels at runtime without extra work or setup.
Of course, that’s a pretty big “if”, but a simple unified system sounds like a great thing. And of course there’s more to this than swapping kernels.
Apertus was developed with due consideration to Swiss data protection laws, Swiss copyright laws, and the transparency obligations under the EU AI Act. Particular attention has been paid to data integrity and ethical standards: the training corpus builds only on data which is publicly available. It is filtered to respect machine-readable opt-out requests from websites, even retroactively, and to remove personal data, and other undesired content before training begins.
We probably won’t get better, but sounds like it’s still being trained on scraped data unless you explicitly opt out, including anything that may be getting mirrored by third parties that don’t opt out. Also, they can remove data from the training material retroactively… But presumably won’t be retraining the model from scratch, which means it will still have that in their weights, and the official weights will still have a potential advantage on models trained later on their training data.
From the license:
SNAI will regularly provide a file with hash values for download which you can apply as an output filter to your use of our Apertus LLM. The file reflects data protection deletion requests which have been addressed to SNAI as the developer of the Apertus LLM. It allows you to remove Personal Data contained in the model output.
Oof, so they’re basically passing on data protection deletion requests to the users and telling them all to respectfully account for them.
They also claim “open data”, but I’m having trouble finding the actual training data, only the “Training data reconstruction scripts”…
It’s “of out hot” and “eat the food” - if we interpret “of in” (derived from “oven”) to be putting in the food, then “of out” would mean to take the food out, and thus “of out hot” would mean you take it out hot, and eat it
Yes, but it doesn’t change the fact that some people won’t use it for that reason, and makes it a fair question if the new model is also “slow”.
I don’t know about making fun of a dialect, but it’s not quite utter nonsense - “oven” sounds like “of in”, so it can be interpreted to mean that it shouldn’t be called oven, because when you put the food in it’s cold, you only eat it when taking the food out, when it’s hot.
The sentence structure is so absurdly wrong it makes me wonder if somebody was genuinely trying to make a pun and ended up with that, or if it was intentionally butchered.
IIRC he was telling people to spam sponsorblock with fake segments, because in some videos he had parts where he just stopped discussing the topic and talked to his (sickly?) cat when it came by, and people were accurately marking those segments as offtopic or whatever it’s called. He was presenting it as evil sponsorblock blocking his sick elderly cat. I could try to find a link later when I’m at my PC.
Other than that, he’s generally alarmist and arrogant, but I do appreciate him and what he’s doing.