That's not how it works. Of course an expert knows stuff, but to be trustworthy they have to prove how they know it. If it comes from literature, then that's a reference. If it comes from their own research, the research workflow mandates they publish their research along with detailed info on how it was conducted to allow their peers to replicate. Once you've done that, they have a reference to use. Which also means if they made a mistake, someone else can prove their mistake and publish a counterclaim.

Because MacOS it's not a BSD. Linux in Android is still open source and still Linux. MacOS might have started from BSD code but it's not BSD licensed, it's a closed source kernel and operating system that was once based on BSD. That actually makes the point that Linux would now be relegated to a small niche if not for the GPL: as with BSD, a corporation could have produced a closed source version of it and dominated the market with it. Claiming it was originally based on Linux wouldn't give us the source code for it...

So you enter your car, set the location, then you go to sleep and the car awakes you at the destination? Because with "Full" self driving, that's the expectation. Anything less is partial self driving.

The removal from maintainers file was not based on russian nationality but on being employed by a company in a list of sanctioned companies https://lore.kernel.org/all/7e... Also this is the offending Linus post https://lore.kernel.org/all/CA.... While he did clearly side polically against Russia, but I don't see him referring to the Russians, and moving against a population is a key ingredient in racism.

Basically all models currently released in the open include the weights, exactly to allow people to run inference on them. The differences among so-called "open" models are usually 1) who can run it 2) what can they be run for 3) is modification and/or redistribution allowed

Copyright does not work that way, you are confusing it with trademarks or patents. Anything creative anyone produces is protected by copyright automatically without needing any kind of registration. Requiring registration would be worse, because then only corporations would have time and resources to get copyright protection. Of course if a copyright case goes to court you have to prove you are the owner of that work, but not by registration.

According to what he wrote in the github issue, there was no email exchange. He got the review, and went straight to pulling the extension from AMO without even trying to reach out to the reviewers. I do not know if he has an history of issues with the review process, but in the github issue he does not say anything about previous incidents. Also, he says he discontinued the Firefox version entirely, the next version will be the last.

In the EU, warranty is mandated by law to be at least 2 years.

As usual, this is clickbait. Nowhere in the original blog post it's said it was unexpected. You run some code, and it breaks because you passed a timeout? the more braindead way to fix that is by extending the timeout, of course the LLM did that. It's not like the LLM was given constraints on what it could edit.

It's unlikely they will, considering Firefox on Android does currently have extensions so they have nothing to apologise for anymore.

actually, you are breaking privacy laws just by storing content without having your users sign a privacy policy; if you are allowing read requests as well you are responsible for the content those users upload to your bucket (e.g. copyright infringment). If you publish any of that, you are again violating privacy laws and possibly copyright (if you get a copyrighted work without a license, there might be some things you are allowed to do by law; redistributing is most certainly not one of them)

> he said that Minsky would have had no reason to believe that she was being coerced Except he didn't say that. From your own link, he just said that nobody presented *evidence* of him knowing the girl was being coerced. It's a subtle difference, but I think he was mainly upset with the technical fact that journalist used the word assault when even the victim didn't claim Minsky assaulted her, but that instead someone else directed her.

You have to accept that those are not US social media, they are global social media. Most of their user base and therefore their profit is outside of the US, even if they were founded in the US. Also, I think all or most of social media company (or any big enough corporation for that matter) are traded in public stock markets where anyone in the world can buy shares of ownership at any time. US created globalization, now you have to live with it.

The point here is not who can do better with less core. The point is we have a 6000â budget. Do I get a faster build with a Threadripper, an Intel Alder Lake, or an M1 Ultra? Well, at multithreaded workloads, we all know the answer to that, even without waiting for Zen 3 Threadrippers to come to market. Intel probably wins at single core performance, and when you add a 3090 or 3080 to both systems Apple ends up possibly winning at power consumption, except they are marketing it as a performance champion.

With that pricing and number of cores, I think a system mounting an AMD Threadripper would be a fairer adversary on the CPU benchmarks. With a 6000â budget you should be able to pair the one with 24 cores with a 3080 and lots of RAM and see what's what. That Alienware Aurora is half the price, FFS...