CamperBob2
CamperBob2 on Hacker News
For whatever requirements are specified by gwerbin.
In two years from now, if I ask GPT-8 or whatever to build me a Web server, what web framework is it going to use?My guess is, you won't care.
How will it and/or I know which one is a good choice?
My guess is, it will make a good choice.
If I'm replaced, it will mean that humanity as a whole has taken a step forward that's normally associated with digging up monoliths. If nothing else, it'll mean that I'll get to do something more interesting.
I'm one of those "code to live" types, as opposed to someone who "lives to code." More specifically, I usually write code in the course of building something that uses it. That's true 99% of the time, anyway. Why would I complain about a technology that helps me (and my customers and employers) move closer to the actual goal?
As for resulting in my death, AI is not going to be what kills me. Not unless humans misuse it. Which, of course, we will, because that's what we do... but (shrug) if it's misused in a way that kills me, at least it'll be a more interesting death. More likely, it will extend my life by accelerating medical progress the way it's starting to accelerate progress in math.
In any case, I don't see the current crop of frontier labs being especially important in the grand scheme of things. Unless they have something up their sleeves besides serving LLM tokens, they are most likely the pets.com and Webvan of our day, not the Amazon and Apple. They are valuable partners, certainly, but I can get by without them.
Yes, I certainly hope so.
And I'll know about theirs. Fine by me.
I have never understood and will never understand why so many programming luminaries fell in love with OpenGL. It violates the most important engineering precept that I try to follow when I design a library: minimize hidden state. OpenGL is nothing but hidden state. You have no way to tell what a piece of code is going to render without seeing everything that set up the stack beforehand, perhaps thousands of lines away, or in entirely different modules that you might not even have access to.
"Man is a thing that will be surpassed. What have you done to surpass him? Wait, no, not like that"
The prereqs aren't as extensive as people fear, but they
are important. Starting out by building a scratch-trained GPT3-scale model is always an option, but anyone who's going to succeed at that doesn't need to watch any YouTube videos.
For the rest of us, the sense of discovery that we get from training and running toy models is a big part of the reward. This stuff is not an exact science, at least not yet, so time spent building intuition with small-scale models is an incredibly important part of the learning process. Not pointless at all.
But machine learning stuff feels like it could have room for learning-by-doing applications that aren't entirely pointless, so I wonder if there's some way to approach the topic I'm simply not finding.That's what Karpathy's videos are for. You end up with a GPT-2.0 equivalent if you follow them to the end.
Meanwhile, on this planet, more billionaires supported Harris in the 2024 election than supported Trump.
Let's see... AI versus the people who thought Donald Trump should be re-elected as President.
Hmm. Let's give the clanker a try.
If you're truly going to build your own model (and commit to the cost of that), then video tutorials will not cut it.That makes no sense. It's like saying if you want an engineering degree, going to lectures will not cut it. Of course not, you have to do the homework and projects too.
There will likely be requirements like registering with the governmentSadly, my GPUs were lost in a tragic boating accident.
When you ask what time it is, the answer ultimately comes from an atomic clock that uses quantum state selection. You could argue that any sort of fluorescent lamp or LED is quantum-driven, based on bandgap thresholds or ionization of gas. Same for anything involving solar PV electricity generation.
The example the author gives, transistors that somehow rely on quantum tunneling to produce gain, is not a good example at all. In fact it's downright wrong. The essay could have been titled "I don't understand transistors."
If an author doesn’t have time to write. Why should I give up my time to read?Because -- if the author is to be believed -- essentially no one bothered to look at this work at all before AI became available to help with it. I had certainly never heard of it before. Had you?
That's the choice being offered. Not between "LLM slop" or "painstaking human authorship", but between something being done or nothing at all being done.
GLM 5.3 and GLM 5.3 Flash are very different animals. The full GLM 5.3 model is what HF used to perform forensics on the OpenAI hack.
Nothing particular special about 5.3 Flash -- you might as well run DS 4.1 Flash at this point, as it's better in every way. But there is as yet no all-around replacement for the full 5.3 model.
I just bought a couple of Reolink RLC-810A cameras, and realized too late (after receiving them) that they require an app just to enable communications and configure the baseline RTSP functionality. Claude found
https://github.com/TinKurbatoff/reolink-init# and took care of the issue nicely.
I will probably buy a couple more of the RLC-810As, since I can only assume that newer models will be harder for the owner to actually own.
I've always said (and been duly downvoted for it), "Don't want people to take it and reuse it? Don't put it on the Internet."
Sadly, being right about something like that does not, in fact, automatically entitle me to founders' shares in OpenAI or Anthropic. But I'll keep preaching it, nonetheless.
I shudder to think of the paperwork the EU would require for authorization to transport antimatter off-campus. But they have hauled it around for up to 20 minutes (
https://www.antimatter.hhu.de/en/about-us/base-step) so in principle it could go a bit farther!
Thanks for the data-centric perspective. It reinforces my vague impression that DS 4.1 Flash eclipses almost everything out there, or at least ties it. We're really lucky to have that one.
(Shrug) HN isn't a place where ignorance should be celebrated; there are plenty of other venues for that.
If the stockholders are OK with it, who's to say otherwise?
I do think it would be a good idea to look at additional regulation regarding the public companies that pension and major index funds can invest in. I'm not crazy about being forced to own Musk properties just because I hold an S&P index, for example.
I don't know; it worked OK when I checked earlier (private browsing with Firefox).
Probably better if no one else clicks on it. I can't edit the post any longer. It won't let me flag my own post, and that might be a good idea too, given what's been going on lately.
I think troupo says it better than I did: "I want people to stop pretending they're surprised when a computer does exactly what it's instructed to do."
(I will agree that the snarky part is better aimed at Anthropic than at you guys, though.)
"Gasoline is dangerous, and nobody but us should be allowed to sell it. To prove this, we put a cup of it in the microwave, hit 'Start', and sprinted away. The results, we believe, speak for themselves."
Wait, is it a death cult or a sex cult?
You know, just in case the distinction proves important at some point.