danbruc

42 ms·

691.
by danbruc 3y ago
Instead, Shell—a massive oil company and one of the many entities directly responsible for destroying our planet—wants you all to know just how rad its fossil fuel products are, and even made a whole Fortnite world for you to enjoy! Sure,
692.
by danbruc 3y ago
I fail to see the relation, what has less wealthy nations catching up reducing wealth inequality to do with the survival of humanity and arms trades?
693.
by danbruc 3y ago
Does it matter? Unless you are suggesting that we should actively try to maintain a wealth differential across the globe, it will just be what it will be. And I do not think that a single factor will be deciding, there will be most likely v
694.
by danbruc 3y ago
Truth-preserving in this context means that the function value is true if all function arguments are true. If you only have truth-preserving functions available, then you can not output false if all inputs are true, hence you can not build
695.
by danbruc 3y ago
Beating the West seems to have quite a negative connotation. Why not just be happy that the rest of the world is catching up and that wealth differences are hopefully shrinking?
696.
by danbruc 3y ago
Or to stick with the software metaphor, you do not have a specification for real universe, so you can not tell whether the implementation is correct or has bugs. What you consider a potential bug might as well be a feature, impossible to te
697.
by danbruc 3y ago
A bit nitpicking. I do not think it is quite right to say that current large language models learn, we infuse them with knowledge. On the one hand it is almost just a technicality that the usage of large language models and the training pro
698.
by danbruc 3y ago
No. Because you have never seen a universe of which you are sure that it was a real universe, so you do not know how a real universe behaves, so you can not determine whether our universe behaves differently than a real universe. You are ju
699.
by danbruc 3y ago
Yes, the recent popularity stems from Bostrom's simulation argument trilemma. But everyone always jumps over that, takes it as granted that we live in a simulation, and proposes some wild ideas how we could find evidence or what might
700.
by danbruc 3y ago
Correct. c - v = 1 cm / 215 ky About one nanometer per week in case that is more relatable.
701.
by danbruc 3y ago
Any reasoning you do about the simulation hypothesis has one gigantic problem - the only thing you know anything about is the universe we observe. You have never seen a real universe and you have never seen a simulated universe, so you have
702.
by danbruc 3y ago
You can not even determine that you are in a simulation if the simulator does not want you to know.
703.
by danbruc 3y ago
If you really think about the simulation hypothesis a bit, you will realize that it is a stupid idea.
704.
by danbruc 3y ago
List the digit sum mod 7 of the first 10 primes larger than 53. This starts with 0 0 6 1. What do you have to do to generate the next number? Count the numbers in the output to figure out that you have done 4 numbers, add 1 because the 5th
705.
by danbruc 3y ago
Yes and I'm saying even that doesn't work as robustly as CoT across different tasks. What does not work? CoT is so obviously not just about padding for extra compute. In a certain way it is but the crucial point is that an aut
706.
by danbruc 3y ago
I did not read the entire linked article but For instance, I tested whether asking a model to produce "a b c d e" before solving a math problem improves performance. is of course not what you would expect to work. It is not abou
707.
by danbruc 3y ago
It does. Those models are glorified look-up tables, output = answers[input]. It would of course be insane to try to store this mapping explicitly but it is still there implicitly in the weight of the network. But no matter what, it is just
708.
by danbruc 3y ago
But 50 % slower is either 1.5 or twice as long, depending on whether the 50 % refer to an increase of the runtime or a decrease of throughput. 200 % slower is only unambiguous because the throughput can not decrease by more than 100 %.
709.
by danbruc 3y ago
It is actually correct, it is slower by 0.5-2x in terms of runtime resulting in a runtime of 1.5-3x. Admittedly it would probably have been less confusing to say that the runtime increased by 0.5-2x to 1.5-3x because slower is more intuitiv
710.
by danbruc 3y ago
Just look up what the intended meaning is, there must be drafts and revisions and discussions from the time the law was created, right? And then apply the intended meaning unless the affected party can make a reasonable case that they were
711.
by danbruc 3y ago
Bringing the water in that pool into orbit will set you back a couple billion dollars.
712.
by danbruc 3y ago
Transactional memory [1], there are existing hardware and software implementations. [1] https://en.wikipedia.org/wiki/Transactional_memory
713.
by danbruc 3y ago
You could also have it in O(1), just store and maintain it as you usually store the length in bytes or code units. If you had all your string operations like substring work with grapheme clusters by default, which might arguably make sense
714.
by danbruc 3y ago
Four. Obviously. The more interesting question is whether the Unicode rules actually give that answer. EDIT: Just checked it using the first online tool [1] that came up and it indeed says four. So all is good. [1] https://online
715.
by danbruc 3y ago
Why would you have to track all the things? The entire idea of the ad based system is that you still get access to the content even if you are not paying for it. There is nothing to be tracked all the way through the system. If I see an ad
716.
by danbruc 3y ago
I could not believe this, so I looked it up. What I found was that in 2015 some Chinese minister complained that Chinese ball point pens were not on par with respect to writing quality when compared with non-Chinese ones. In 2017 they had t
717.
by danbruc 3y ago
Was this not already pointed out in one of the theoretical papers during the initial replication attempt including the judgment that the interesting crystal structure is unlikely to be producable because it is not energetically favored?
718.
by danbruc 3y ago
The companies and even more so their management are just intermediaries. Oil companies are not producing oil for the sake of it or to raise carbon dioxide levels, they are doing it because people buy the stuff made from oil and they can mak
719.
by danbruc 3y ago
Nobody should care what the boss of an oil company wants to do and hope he wants to do the right thing and hope he will not change his mind in the future. Just force them to do the right thing, make oil less profitable if that is what you t
720.
by danbruc 3y ago
One and a half decades actually.

More ›