pona-a

19 ms·

271.
by pona-a 1y ago
I see a common thread in that comment: there's too many distros, there's too many languages, there's too much software. But there's no central authority telling people to make them. You can't force people to not rei
272.
by pona-a 1y ago
> Every Linux distribution is different A distribution is just a collection of software to handle common needs. Most are quite similar: systemd, coreutils, glibc, dbus, polkit, pipewire/pulseaudio, and a DE, typically GNOME or KDE.
273.
by pona-a 1y ago
vector product
274.
by pona-a 1y ago
Game theory?
275.
by pona-a 2y ago
How so? I'm writing this from an Fedora Sericea, which is Silverblue but with Sway instead of GNOME. Atomic Fedoras solve only package hysteresis (your package manager being unable to reproduce the intended system state because of unac
276.
by pona-a 2y ago
Most regulations are motivated by universal normative prescriptions. People don't want to be served asbestos in their food, structurally unsound homes being built, or data brokers conspiring to raise your insurance, regardless of the v
277.
by pona-a 2y ago
Previous discussion: 31 points | 22 hours ago | 8 comments https://news.ycombinator.com/item?id=43746981
278.
by pona-a 2y ago
This kind of nonsense prose has "AI" written all over it. In either case, be it if your writing was AI generated/edited or if you put so little thought into it, it reads as such, doesn't show give its author any favor.
279.
by pona-a 2y ago
There's Cloudflare PrivacyPass that became an IETF standard [0], but it's rather weird, and the reference implementation is a bug nest. [0] https://datatracker.ietf.org/wg/privacypass/about/
280.
Gemini tries to solve a Portal 2 test chamber [video](youtube.com)
1 points by pona-a 2y ago|0 comments
281.
by pona-a 2y ago
Water fluoridation conspiracy theories were a thing old enough to be a plot point in 1964 Dr. Strangelove. When there's an entire grift or political career to be a "Fluoride truther", it's really hard to take seriously a
282.
by pona-a 2y ago
> Giving 10 month old children permanent crutches may harm the development of walking, but since the crutches make unassisted walking obsolete, maybe it wasn't worth it anyway? LLMs are a language calculator, yes, but don't sha
283.
by pona-a 2y ago
Detecting the actual AI generated content is not an easy problem. Not following deep links and recognizing the particular website template and structure is easier. I really feel a monoculture of anti-bot tools can defeat their effectiveness
284.
by pona-a 2y ago
Volume and incentives. Before this LLM craze, the biggest crawlers were search engines. They had a motivation to not bring down their targets, because who needs an index full of dead links. With LLM crawlers, all you need is text, and if th
285.
by pona-a 2y ago
A lot of tokenization bugs came from odd Reddit subreddits, such as r/counting. And back then, we didn't have free model weights, which is they all converge to similar weights, might tell us a lot about how to design adversarial e
286.
by pona-a 2y ago
It just feels strange to muse about the embodied experience of drawing while having a computer spit out a semantic average of your general idea with little creative control.
287.
by pona-a 2y ago
It seems from the finer details, a lot might be AI generated, which at least to me would defeat the message about how our interfaces are less human. I'd be very glad to be proven wrong, and it's just the style.
288.
by pona-a 2y ago
> Why reinvent the wheel when we already have Reticulum network Assuming README is among the first files created in a project, here's the date of the first commit for each: markqvist/Reticulum: Apr 4, 2018 briar/bria
289.
by pona-a 2y ago
You are facing a constitutional crisis right now, something most EU members can say they do not. > Khalil called his lawyer, Amy Greer, from the building's lobby. She spoke over the phone with one of the ICE agents, who told her
290.
by pona-a 2y ago
I wrote a tiny pipeline to check, and it seems styled Unicode has a very modest effect on an LLM's ability to understand text. This doesn't mean it has no effect in training, but it's not unreasonable to think with a wider co
291.
by pona-a 2y ago
I remember when the first Stable Diffusion models came out, one of their model cards, in the "safety" section, mentioned using the model to generate depictions of historical events. At that time, I thought that warning was about a
292.
by pona-a 2y ago
Can you point to any innovative Windows tiling WMs and explain what "workflow advances" it makes? All I found was FancyWM, and it seems basically identical to i3.
293.
by pona-a 2y ago
I think this idea of "generative UI" stems from a genuine confusion about what Spotlight, Alfred, or Chrome Omnibox are: they are a collection of utilities, not an app of themselves. They work because a designer has put some thoug
294.
by pona-a 2y ago
Math Academy uses spaced repetition for skills with tiny, to-the-point interactive lessons (typically following "theory, some exercises, theory, some more exercises" formula) based on an initial diagnostic test, where the skills a
295.
by pona-a 2y ago
I am currently studying for our country's version of the SAT and, having tried Math Academy — having been convinced there is nothing anywhere as polished and developed on the market — I still had to cancel my subscription after the fir
296.
by pona-a 2y ago
In-context learning is a very different problem from regular prediction. It is quite simple to fit a stationary solution to noisy data, that's just a matter of tuning some parameters with fairly even gradients. In-context learning impl
297.
by pona-a 2y ago
Would that make in-context learning a superset or a subset of pretraining? This paper claimed transformers learn a gradient-descent mesa-optimizer as part of in-context learning, while being guided by the pretraining objective, and as the p
298.
by pona-a 2y ago
> If you posted that challenge to an intelligent person from 20,00 years ago, they would just stare at you blankly. Depending on how you pose it. If I give you a long enough series of ordered cards, you'll on some basic level begin
299.
by pona-a 2y ago
I don't think so. A lot of useful specialized problems are just patterns. Imagine your IDE could take 5 examples of matching strings and produce a regex you can count on working? It doesn't need to know the capital of Togo, metabo
300.
by pona-a 2y ago
I feel like extensive pretraining goes against the spirit of generality. If you can create a general machine that can take 3 examples and synthesize a program that predicts the 4th, you've just solved oracle synthesis. If you train a n

More ›