mlyle

58 ms·

901.
by mlyle 2y ago
I wonder how much of this is an intrinsic limitation of LLMs, and how much that interdisciplinary thinking and mashing together of problem domains is missing in the training data. It's a pretty rare thing, and the only times these ana
902.
by mlyle 2y ago
The discussion here is leading with the idea that there is collusion and monopolization of publishing marketplaces that, in practice, screw consumers and most authors . I do not see anything that "question[s] whether authors and edito
903.
by mlyle 2y ago
I think there's a crucial difference here. If we want to prohibit an activity, having a fine that's the same in dollars is not great because of the whole inconvenience-fee thing. If the purpose of the fine is punishment and de
904.
by mlyle 2y ago
> very few authors really make a living from book proceedings. Sure, but there's very few notable authors, too. > industry that does their best to exploit the authors Granted. There's a lot of creative works that are very,
905.
by mlyle 2y ago
> It doesn't do any controls over the number of copies "out". This is untrue; under normal circumstances it had matching numbers "out" in the commercial book collection. It was madness in early 2020 when they te
906.
by mlyle 2y ago
We're supposed to discuss better than that here -- to discuss the actual disagreement in good faith. People who create stuff should have an exclusive right to publish their work-- for awhile. But not forever; our culture should not en
907.
by mlyle 2y ago
I support the work of the internet archive, but I also think that for a nonprofit in this role, survival is job one. The risky stuff needs to be done via cut-aways that, if defeated, don't wipe out all the other good. And even these cu
908.
by mlyle 2y ago
> getting as much as possible done - be as productive as possible I didn't get this at all as a main thrust of the article. The author doesn't like re-reading the works that impacted him the most, because he feels he loses some
909.
by mlyle 2y ago
There's funny things at play even beyond training effects, most likely. Even leaving aside that finding a dead person might be something that inherently triggers avoidance. People screening luggage for bombs/knives/etc do si
910.
by mlyle 2y ago
To answer you in a different way: Today we would build these systems with FM or digital voice. But there are enough advantages we've gotten used to (graceful degradation, better handling of doubling, etc...) that it's not really u
911.
by mlyle 2y ago
Unfortunately, in practice, this isn't really true. It's a less powerful motivation, in that there are not shareholders who will benefit from growth. But in practice, executives and managers end up making more, which provides an i
912.
by mlyle 2y ago
Not everything beneficial to the world has a viable business model. In the best case, you get to change the world and make a bunch of money. Sometimes, you just have a viable business selling sugar water. And other times, you're do
913.
by mlyle 2y ago
> More precisely, DUV has about 10x shorter wavelength than EUV Other way around, though 10x shorter is confusing; I suggest saying EUV has a wavelength that's about 1/10th of DUV.
914.
by mlyle 2y ago
Those are all situations where a commercial entity has put private funds at risk. Most of these defense contracts that have gone way over have left the taxpayer paying for it all. In many cases, the defense contractor ends up better off
915.
by mlyle 2y ago
So, deciding not to comply with the norms of foreign jurisdictions is a risky business. Aggrieved judges from other jurisdictions can and will do things that will make your life miserable-- seeking to seize foreign assets, filter locally,
916.
by mlyle 2y ago
The thing that's a little tricky here, is in practice you can pretty easily keep a copy of that scan. It makes subsequent infringement a whole lot easier than checking out the book from a library or borrowing a physical book from a fr
917.
by mlyle 2y ago
I'm having a hard time aligning "voluntarily" with "in compliance with a court order" in my mind.
918.
by mlyle 2y ago
Mathematica is great.
919.
by mlyle 2y ago
If I save a little bit of work now, my life is better. In the unlikely event that the system is still used in 76 years, it will not be my problem. I'm generally against rolling problems to future generations, but taking risks with a m
920.
by mlyle 2y ago
The fact that it went so well is not evidence that no original issue existed. On the other hand, maybe it's evidence that we over-invested a bit into diminishing returns. A perfectly fine-tuned response would have a little bit more to
921.
by mlyle 2y ago
> What, ultimately, did Wolfram do in these five most productive years? Can it be said in a sentence, or a paragraph? He shipped a lot of software, wrote a lot of books and papers, did a lot of media, and substantially advanced his passi
922.
by mlyle 2y ago
> Finally, it was a fake job because the problem that the Conglomerate had hired Andersen to solve was not real, I don't really like the whole popular understanding of the y2k thing. Substantial problems with infrastructure were
923.
by mlyle 2y ago
> but distributing data about active criminal investigations is malicious, or at least dangerous. Is showing some reporters some sample data to show that the data exists malicious? Because I believe that's all he's been accus
924.
by mlyle 2y ago
The computer was on its own battery to get clean power. And it was using more than 200mA "power", not that mA is a unit of power. Honestly, it feels like you're trying to prove me wrong about my one case. Power consumption
925.
by mlyle 2y ago
20-30% more battery life and having more performance would have simplified our life, yes.
926.
by mlyle 2y ago
Yup. I had students running a machine learning model on a Pi attached to a scooter. The power draw of the computing system was significant, and the performance limitations were too. The existence of commodity, well-documented, well-suppor
927.
by mlyle 2y ago
Pi 5's peak consumption is higher, but if you are doing the same amount of computational work (but a bit above idle) it is less. Now this uses even less, and even the idle powers are comparable to 4. This wins in every way over Pi4, e
928.
by mlyle 2y ago
> it seems like even simple techniques could extract a lot of signal from the noise, which is just too difficult with the tiny number of low sample, high variance RCT studies we currently use. On the other hand, RCTs gather relatively we
929.
by mlyle 2y ago
> Personally, I have never seen a case where tattoos improved someone's health My dad had small fiducials tattooed onto his body to index where radiation was being delivered, which resulted in the cure of his prostate cancer. Medica
930.
by mlyle 2y ago
Haha. Your uncle probably had waitstaff that did not understand that he was talking about a tax distinction and thought he was creepy. :P

More ›