Isn't citing encyclopedias shunned in academia in the first place? Wikipedia is an encyclopedia and should be treated as such. I understand that access to primary sources can be difficult at times, but in that case, wouldn't an argument based on a citation from an encyclopedia have a shaky foundation?
Fun trivia: the greeting only started to become a thing during the Meiji era as part of moral and ethics education. It may be argued that it has its roots in Buddhist teachings, but there's a theory that the drive to make it a common household practice may have come from the envoys who came back from western countries and saw the Christian prayer and thought it was a good thing to imitate. It's quite a stretch, but personally I find it quite interesting given that it was an era where Westernization was the craze.
The majority of anime, at least the Japanese ones, do resemble each other a lot at times, but the AI anime style resembles none of them. It has this distinct style that makes it recognizable as AI art. It is more similar to the style used in illustrations, like the ones you see in many gacha games. And the same style is used even for AI generated comics and illustrations. I think that was the grandparent poster was trying to say when writing "consistent in every medium".
That said, AI is pretty good for creating goofy stuff that otherwise won't sell to the mainstream audiences. My recent favourite AI anime is Chemical Cooker Ryo.
The last time I heard HR being completely let go was with a fintech company Bolt. Then again, that company was midsized, around 200-500 people or so. For larger companies, it's going to be difficult to even realize that HR is redundant in the first place.
I think that's because it's easier to package a DOSbox program than a Win16 program (it had Win32 edition, but supposedly the installer itself was Win16). According to pcgamingwiki, there's a patch to convert the GOG DOS edition to the Win32 edition.
It'd be great if IDEs can store a stack of functions currently being explored similar to what you get when debugging. Not breadcrumbs, but plain stack. Bonus points if you can store multiple stacks, and give them names according to the context.
We have OnlyOffice as an alternative today. Personally I find the UI quite pleasing. But lately I haven't had any need to use office suite at home so I don't really use it so I have yet to find anything to complain (meanwhile, LibreOffice was horrible, while Office 365 was bearable until you stuff too many things in the equation editor).
From my understanding, the code is MIT, but the model isn't? What consitutes a "Software" anyway? Aren't resources like images, sounds and the likes exempt from it (hence, covered by usual copyright unless separately licensed)? If so, in the same vein, an ML model is not part of "Software". By the way, the same prohibition is repeated on the huggingface model card.
That's one thing, the other is you find out you were optimizing for the wrong thing, and now it takes more effort and time to reoptimize for the right thing.
The reason is floating point precision errors, sure, but that check is not going to solve the problems.
Took a difference of two numbers with large exponents, where the result should be algebraically zero but isn't quite numerically? Then this check fails to catch it. Took another difference of two numbers with very small exponents, where the result is not actually algebraically zero? This check says it's zero.
You can add the guideline, but then people would skip the "I asked" part and post the answer straight away. Apart from the obvious LLMesque structure of most of those bot answers, how could you tell if one has crafted the answer so much that it looks like a genuine human answer?
The point still stands. The human body still isn't going change. That's why insulin pump can afford to have all kinds of rigorous engineering, while web-facing infrastructure on the other hand needs to be able to quickly adapt to changes.
The best thing about gccgo is that it is not burdened with the weirdness of golang's calling convention, so the FFI overhead is basically the same as calling an extern function from C/C++. Take a look at [0] and see how bad golang's cgo calling latency compare to C. gccgo is not listed there but from my own testing it's the same as C/C++.
I always install WinCompose specifically for this. Then a Greek letter is as simple as pressing Compose with *, followed by the related Roman letter. Though I still struggle to remember which key maps to which for the letters with no real 1-to-1 mapping (like theta).
Brings back memories. There was a time when the fork, libav, became the default on Ubuntu, and ffmpeg commands would say "this command is no longer maintained" or so. That was where I learned that there was a fork, and I thought ffmpeg was going to die as a result because there was heavy development activity on libav compared to ffmpeg initially. Surprise, ffmpeg outlived its fork!
It's really unintuitive how a mere increase in number of dimensions could totally change the scale of things. This is the crux of the curse of dimensionality in machine learning. If you simply stacked it high, it may almost reach the sun (but no quite), but once you lay it into roughly a square (20000 x 50000), it spans only over a mile per side. To lay it roughly into a cube of 100 dollar stacks, you'd have something like 65 x 150 x 1026, which turns out to be about only 12 yards per side.
EDIT: I screwed up in the 1D calculation. A 10 million-height stack of 100 100$ bills only reaches 67 miles.
Isn't a clock and a gyroscope precisely two things that are needed for the missile to know where it is at all times? Like, it can then know where it isn't by subtracting where it is from where it isn't, or where it isn't from where it is, to get the positional deviation. Combine that with the clock deviation, it can get the velocity and acceleration and then use all three information to generate corrective commands.
The parent comment is not about following examples, but rather that the impact Streisand effect is going to be very limited, and the common folk will not bother to circumvent.
While the search feature is nice, the reference itself still lacks some details about what an instruction actually does. Take for example, [1], and compare it with say [2] (with diagram), [3] (ditto), or [4] (only pseudocode but helpful nonetheless). Of course, all the alternatives mentioned only cater to x86 but still it'd be great if this site also follows the approach taken by the other three.
Wow, it didn't appear when I searched for "Amazon Japan" and filtering for "past week". Now, both appear.
That one was timestamped 2025-04-26T04:39:02 while mine is 2025-04-27T08:42:16, more than 24 hours apart. The search indexing is done less frequent than that?
I don't know why but I get this uncanny feeling when looking at the restored images. Maybe it's because I know it is restored, I wonder if I'd feel the same way if I find it in the wilds.
That is strange, I thought this had always been the case. Long ago, I had a chat about feature selection and dimension reduction, and I expressed my preference of not resorting to PCA because it doesn't actually remove features. After some time, I had a new session chatting about multicollinearity and what adverse effects it might cause, and among the suggestions for addressing that, it suggested dimension reduction, saying "Although you don't prefer PCA ...".