I am not sure it would not help commercial solutions, if all experts are used all the time, sure, but if for example the usage is biased for some experts it would enable one machine to serve more users in parallel or save on VRAM or DRAM without compromising response time, hence cutting costs.
Yes, I believe it was for a Mixture of experts model, which just goes to show how naive IA implementations are at this point, The cache was not solving any hard problem and yet for some reason it was not only not already standard practice but also somehow a notable achievement.
The other day I saw a video talking about this new innovation on LLM inference side of things where they keep some more used weights in RAM and others less used on disk. I always suspected from the sample code I stumbled upon on the IA world that should be extreme opportunities for optimizations. But I cannot stress it enough how dumb the LLM world is where the basics of implementing an LRU cache is pass of as some big innovation. Like any half competent comp-sci or comp-eng professional know about the basics of mitigating this basics bottlenecks like “the data does not fit on available RAM”, “The disk is slow”, etc.
So is not surprising that now that it seems that the “powerfulness” of this LLMs is starting to plateau that we would start to see some improvement in performance/resource utilization and hence running costs.
Thank you for your detailed comment.
Unironically the part talking about song lengths and types of showers was really funny, now I keep thinking in characterizing things in terms of song lengths. It opens so many possibilities of how to express and do fun jokes. My wife and I already say stuff like “Are you going for number 1 or 2? … Let me use the bathroom first then I need to quick do X” But now the possibilities are endless, like “Are you going for a Blitzgreig Bop or a Stairway to Heaven?”
Anyway, back to the main topic, I remember doing something like that years ago, like timing the shower or using a playlist to gauge how long I am taking, but I guess nothing that detailed.
I feel like your breakdown maybe leaving some parts behind or take different time for me, like I scrub my ears and because of my piercings it takes a little more time. I also like to scrub between my toes and that is kinda fast but adds up. But I guess for me the true challenge is the same of comics someone else shared in this thread, really hard not to stop for a minute and take a deep breath, feel the hot water and just enjoy it for a while and gather my thoughts and energy, not to start the day for me, but to go to bed.
Anyway, much rumbling about nothing on my part, but I wanted to share, mostly to say I really enjoyed your comment and the time and effort you put into It. At least for me I usually dress leaving long comments because I am usually on mobile and hate typing on the phone.
I know, that is me. Very short and each day less and less hair. Maybe I’m just slow or clumsy ¯\_(ツ)_/¯
I am sorry, one song? 1, uno? That is crazy for me. I am all for quick showers but unless you are like a Pink Floyd fan I would be running like crazy to finish a shower in less than 5min, unless you don’t wash your hair every day which I do but I know for some types of hair people don’t.
That is not the purpose of Secure Boot. The purpose is to establish a chain of trust of all code running on the system from boot and as such eliminate rootkits that can hide from the OS. A classic example is the MBR bootkit. Of course that this is mostly out of the window if at some point in the chain the trusted code just runs untrusted code, like the bootloader or the OS running unsigned code. Also the implementation is terrible, a proper implementation would allow the user to use their own certificates and only their own certificates, otherwise a compromised generic certificate fucks everything up like it already happened.
Public transit is common in my 3rd world country and I was surprised how more unsafe it feels in the US. My theory is that it is a by product of the car centric society. For one there is safety in numbers and because it is so underused any one person is more exposed. But also because almost the only people that take the public transit is people who cannot drive or be driven for some way or the other it inevitably have a bigger proportion of mentally unstable people for example.
People it this thread does not know what the Turing test actually tests. It is Ok, I guess the blame fails on pop culture.
The Turing test is not a test of sentience, it is a test if a machine can fool a human being into thinking it is sentient. And machines have passed this test for decades already. Who would have thought that our predisposition to anthropomorphize everything would make us so susceptible to thinking stuff is sentient 🤷
It is. There is limitations to a Turing machine that a human brain does not have. I expand on that on this other comment here.
There’s nothing to indicate the human brain can’t be implemented in a Turing machine too.
That is not true, this is a research branch from computer science and math, it is called Computability theory, it deals with the limits of expressiveness for different types of theoretical machines and expressions, and the most expressive of all is the Turing machine, and a Turing machine cannot do some stuff, the classic example is the Halting problem, a computer cannot definitely say if an algorithm ever stops (mathematica proven that it cannot do it), but a human can do so quite easily.
One may think that maybe a Turing Machine cannot do something but can simulate another machine that does, but that is also proven to be impossible, it cannot simulate something more expressive than itself.
I will have to defend users here and blame AI scientist/developers together with the whole docker culture. The way they package and delivery software is just terrible. And even though pytourch has support for all GPU apis the devs seems to have a hard time adding this support. Granted, part of that is also on pytourch that for some reason Don’t seem to be able to make a unified release that supports all GPUs and then you need different versions and by default IIRC they default package only supprt CUDA and CPU.
deleted by creator
The only way to address systemic problems like crime and violence in any meaningful way is addressing the underline problem that pushed people to commit crimes. That underline problem is quality of life which is directly related to income.
Taking people out of the informal job market is another problem to tackle that can be helped with having better work laws. Having good working laws stimulate people to get jobs in the formal market and avoid offers for informal work even if with better pay. That is not a problem that would be solved in the short term but this laws will help with that. The same approach was used in other countries and together with more government enforcement and fiscalizaton of the companies you see great improvement over the years and decades migrating people to the formal job market and reduction in crimes and violence.
It is worth pointing out that the only reason docx support is lackluster is because of active efforts from MS to undermine third party support. It is a interesting story actually, but I don’t recall exactly, but it goes like this, there was some regulatory push to open formats and MS undermined this by creating and making the docx (and all other *x family) open but make it so convoluted and unnecessarily difficult. If I am not mistaken they even keep unnecessarily updating and changing it so third party is always lagging behind in support and the cherry on top that MS don’t correctly implement its own format on purpose so the files are effectively broken for anyone that follows the spec to a T.
I read about that a couple of years ago, I am speechless that such a thing is even allowed to exist.
Definitely the latter. I unfortunately use a lot of Google services and they are pushing AI integration in everything and making it very hard to opt out or they keep pesting you to enable it.
I “love” when the wrong technology is applied.
On the the dawn of the smartphones Mozilla tried to enter the space with an FirefoxOS and the pitch was that every app was just a website just more tightly integrated with the phone. The problem is that all the web stack is wonderfully resource hog and at the time phones were super underpowered running websites were not optimized in a browser that were not as optimized as today. So it was a terrible choice for the time being.
Other good one was Android early days. They choose Java as the default app environment and development. It kinda makes sense to use it if you want the same program to run on different platforms, the problem, again, it runs worst and with the underpowered devices of the time everything was a slog. And they doubled down on the mistake by using a garbage collector that doubled the memory usage of every app. The cherry on top, at least in hindsight is that arm was and still is the de facto Android plataform, greatly disminishing the advantage of using Java/JVM. And today Google enabled apps with native code optimized for specific plataforms, but everyone only care about ARM so of you try to run Android like in an Intel laptop a lot of apps are not compatible.
End of rant.
This reminded me of a joke patch someone sent to Firefox to rename incognito mode to porn mode.
I bet those are the same kind of people who copy paste
curl www.trust.me/bro.sh | bash -on their terminal. No respected sw dev professional would run unsupervised random commands on their machine just as much they would not send unsupervised random requests on the open Internet.These AI people are fucking morons.