My point is that the essence, the core mechanism, is typically a 1-liner, but only with a pedagogical context, not a production (in the sense of anybody else using it reliably in realistic situations) context.
1,986 karma · joined December 9, 2014
My point is that the essence, the core mechanism, is typically a 1-liner, but only with a pedagogical context, not a production (in the sense of anybody else using it reliably in realistic situations) context.
I do have non sudoers groups devices and I also have shared devices with non sudoers users. It works. It's not necessarily for everyone, e.g. not for powerusers who want to go fast very often, but it's definitely usable for most users still.
well that's the trick. If you look at implementing a Web browser or a physics engine you instantly give up, and you are right about it, it's just too hard. If instead you write a partial HTTP client which can display some text from a URL or have a 3D scene and in it an object Y coordinates get smaller each millisecond then suddenly in terms of the essence of each, you are 80% there really. Now if you want to implement the "rest" you might need an entire lifetime, or more. Still IMHO it's definitely worth doing very compact and limited versions of all those in order to learn, even if done only once in your life.
- code gets re-regenerated over and over again, so waste of resources
- no signal back upstream that it is indeed a genuine need that is in fact so important some users did dedicate resources for it, maybe more users would benefit from it
FWIW houses in Tuscany with very strict rules, e.g. you don't even get to pick the color of your own house, manage to get the entire place setup with full-size AC it's feasible. What's not feasible is trying to ignore laws on what a building must look like but that has nothing to do with air conditioning. For example if Belgium if you has is classified as notable (most old houses are) then you need a permit even for the window frame.
Proof in Belgium : https://www.coolblue.be/en/air-conditioners/filter which ironically enough aren't even available anymore because they are out of stock, so clearly purchased showing people had no problem getting them. Alternative example 12000BTU A+++ energy rating delivery next week https://www.bol.com/be/fr/p/climatiseur-mobile-puissant-clim... To be clear I'm not advocating to buy these, or not, only that they are freely and conveniently available.
Well it did work, in the sense that I managed to complete some tasks. I don't think I ever collected the money/credits back then though (simply because it amounted to so little). What I can attest though is that... it was debilitating. If you think your office job is boring then splitting it in way smaller tasks where you have no autonomy is absolutely terrible from a worker standpoint.
I initially was hoping to use it in order to work on providing a service over a dataset but understanding first hand what it takes to make it happen made me stop. It radically changed how I saw supervised learning since, and sadly not in a good way.
What did they have though? Optic fiber, high bandwidth, low latency, reliable and for a very low price.
Honestly I think ADSL was already enough for most people, I mean non professionals, but now I think France has the infrastructure for a good enough Internet for the next few decades. I can't fathom normal usages that would need higher bandwidth and lower latency.
The author is most subtle or diplomatic than me and thus call them "AI vendor press releases" or "AI vendor marketing" but we know.
Can, but do they? Feels a bit like Musk's Colossus that "could" use nuclear power but has been running on "temporary" natural-gas turbines for years now.
I do think it is an interesting exploration but the jagged frontier makes it really challenging to know what will consistently work and what will not, to the point that I bet (maybe pessimistically) one will be gradually less daring with creative plan with their "companion" simply because they can't trust it.
You naively thought Silicon Valley was just a funny TV show about people wanting to make money, a lot of it, but even that was an idealized version.
I mean first that is already what plenty of 17yo are actually doing, because that is what they do at school or in parascholar activities. There are already countless of such tutorials where you can do that in an afternoon.
The pointless part though is precisely why Amazon and others are hunting for rare books, all the low hanging fruits have been picked already so just training a bigger model will simply mean burning more energy and money. Sure training a small one for the basic principle is a great pedagogical thing, training another one, medium, then maybe a large one, is also good in term of learning the process and architecture, but one should not expect it to be useful out of that context.
Pure players are precisely doing everything they can to corner the market by making their own scale unreachable by others. Smaller players with access to lesser infrastructure are thus betting on different market, e.g. embedded systems.
17yos should definitely build their (L)LMs from scratch and whatever bigger model they can train for free, or for cheap, but they should not expect that to bring them any riches.
my keyboard journey is on https://fabien.benetou.fr/Tools/Keyboards