loudmax

11 ms·

121.
by loudmax 1y ago
Most of these models have been trained using 16-bit weights. So a 1 billion parameter model takes up 2 gigabytes. In practice, models can be quantized to smaller weights for inference. Usually, the performance loss going from 16 bit weigh
122.
by loudmax 1y ago
Fellow Virginia Tech alumnus here. "VA Tech" is perfectly legible and unambiguous. "VT scientists" is straightforward to Americans, but not everyone in the world will know what "VT" is in this context. It co
123.
by loudmax 1y ago
As a system administrator, I vastly prefer to deploy Go programs over Java programs. Go programs are typically distributed as a single executable file with no reliance on external libraries. I can usually run `./program -h` and it te
124.
by loudmax 1y ago
My theoretical explanation for the gap is that women are sensitive to much more than physical appearances when selecting a mate. When men select a mate, they over-index on physical characteristics, relative to women. So, women will take in
125.
by loudmax 1y ago
It's the Latent Space podcast, published June 13. It's a pretty decent podcast for keeping on top of AI stuff. Swyx, one of the co-hosts, is active here on HN.
126.
by loudmax 1y ago
Programmers aren't deterministic either. If I ask ten programmers to come up with a solution to the same problem, I'm not likely to get ten identical copies. Different programmers, even competent experienced programmers, might h
127.
by loudmax 1y ago
There is a proposal to build an apartment complex in my neighborhood in Northern Virginia. The houses that had Trump lawn signs up last year, now have lawn signs arguing to block the new housing (or "preserve the neighborhood's ch
128.
by loudmax 1y ago
As a pro-nuclear-power fan, this does sound like it could be good news. I do have a few concerns. First, the power source is orthogonal to AI. Datacenters will run just as well on nuclear, or coal, or wind and solar. And the electricity
129.
by loudmax 1y ago
I'm afraid a lot of Americans consume the news like they consume sports media. They root for their team and select a news stream that presents them with the most favorable coverage.
130.
by loudmax 1y ago
The OP is not equating contextual with insidious. They're pointing out, correctly, that contextual ads can be insidious. And if they're profitable, they probably will be. A lot of the companies offering LLM services are in a
131.
by loudmax 1y ago
The current US copyright duration is 70 years after the life of the author. This is absolutely bonkers. 50 years from publication would be a significant improvement. 50 years ago was 1975. If copyright were limited to 50 years, we'd
132.
by loudmax 1y ago
Yes, but do artists have the ability to actually monitor and enforce this? You have to have the capacity and the wherewithal and to test these models to even know that your data is being ingested into AI. Big companies like the New York Ti
133.
by loudmax 1y ago
The amount of exposure matters here. It's one thing for a teenager to manage to occasionally access clandestine pornography (as was kind of the norm growing up in the 1980's). It's a much worse thing to spend many hours per
134.
by loudmax 1y ago
When a news organization hires a journalist to file a report, they put their reputation on the line in the hands of the journalist. As a consumer of news, you put your trust in the institution to have a reasonable vetting process, and also
135.
by loudmax 1y ago
I think placing trust in technology is misguided. Bad actors will always figure out a way to abuse the machines. What we can do is place trust in particular institutions, and use technology to verify authenticity. Not verify that what the
136.
by loudmax 1y ago
I don't think I'd want to read a novel that was generated by an algorithm, but I might be up for a Choose Your Own Adventure style game, which might be a better analogy to Minecraft or nethack.
137.
by loudmax 1y ago
Companies that leverage LLMs and AIs to let their employees be more productive will thrive. Companies that try to replace their employees with LLMs and AIs will fail. Unfortunately, all that's in the long run. In the near term, some C
138.
by loudmax 1y ago
Along the same lines, it's often useful to think of an LLM's "understanding" a question or a context. Sometimes responses from smaller models show a shallower "understanding" of the question than you'd ge
139.
by loudmax 1y ago
It can be hard to convey sarcasm, especially in writing, and especially when you don't know the actual stance of the person writing. If I mention "big beautiful tariffs" in a peer group of liberals, they can assume I'm b
140.
by loudmax 1y ago
If the diffusion models are an improvement over autoregression models, then the answer is No, due to Jevon's paradox. That is, as these models get cheaper and better, they provide more utility, driving more demand. Even as your datac
141.
by loudmax 1y ago
The target market for these offerings are corporations, or self-employed developers. If these tools really do make your developers more productive, $250 a month is easily justifiable. That's only $3000 per year, a fraction of the cos
142.
by loudmax 1y ago
This is approximately the premise of the Black Mirror episode "Hang the DJ"
143.
by loudmax 1y ago
Socrates had this to say about literacy: > In fact, [writing] will introduce forgetfulness into the soul of those who learn it: they will not practice using their memory because they will put their trust in writing, which is external and
144.
by loudmax 1y ago
These are conservative values: * The rule of law * Moral integrity * Respect for traditions and institutions There's a tension between those values and other values such as a sense of fairness. People could overlook a polit
145.
by loudmax 1y ago
Viktor Orban's Hungary is illiberal. North Korea is a dictatorship. One is bad, the other is worse. The danger of conflating the two is that you're setting people up for false expectations. If you go around telling people that H
146.
by loudmax 1y ago
That was more or less the case during Trump's first administration. I think a lot of normie Republicans were hoping for a repeat of that. The ones that aren't in denial are being gravely disappointed.
147.
by loudmax 1y ago
The term for this is extortion. The last time tariffs were this high, it led to rampant corruption as companies would pay off customs officers. This was one of the reasons for switching to an income tax. For the current administration this
148.
by loudmax 1y ago
Incidentally, this concept of modifying one's ecosystem comes up in Richard Dawkins' book The Extended Phenotype, which is a follow up to The Selfish Gene. One of the illustrative examples is beavers turning fast moving streams i
149.
by loudmax 1y ago
The "Game Design Otter" action figure seems to come with a pair of flashlights. I bet that's a residue from the previous prompt about illuminating the tablet with a flashlight.
150.
by loudmax 1y ago
That is a very interesting point about how little use of AI most of us making day to day, despite the potential utility that seems to be lurking. I think it just takes time for people and economies to adapt to new technology. Even if techn

More ›