1,049 karma · joined April 21, 2023
Technically, from a geography point of view, euro-using countries as official currency are closer than USD: https://eur-lex.europa.eu/EN/legal-content/glossary/outermos...
(I.e. French guiana's currency is the Euro)
Also how much fentanyl and methamphetamine smoke do you inhale/touch everyday in your car? Compared to public transit
https://www.washington.edu/news/2023/09/07/uw-assessment-fin...
>> Researchers detected methamphetamine in 98% of surface samples and 100% of air samples, while fentanyl was detected in 46% of surface and 25% of air samples.
> WHY? Oh, why?!
> And yes, it turns out that other people have noticed this anomaly. It's screwed up encoding and decoding in their projects, unsurprisingly. We found a (still-open) bug report from 2018, among others
If anyone knows which library/language these issues the author is talking about are in, please tell us. I'd like to avoid that library if possible
Ah that guy. Glad to see some consistency in the world, with stupid ideas/statement coming from same person.
Not really HN worthy post though, flagging.
Naive question: why the need to kill the ssh session? Can't you just open another terminal window or tab (or tmux/screen tab if that's part of your workflow)
Must have something wrong with my setup, but no big deal, for my minimal usage of it, and the amount of time spent, fp32@12s/iter is fine
How fast is it supposed to go?
Just tried on linux with `cmake .. -DGGML_OPENBLAS=ON` on a AMD Ryzen 7 5700g (no discrete GPU, only integrated graphics)
./bin/sd -m ../models/sd-v1-4-ggml-model-f32.bin -p "a lovely cat"
[INFO] stable-diffusion.cpp:2525 - loading model from '../models/sd-v1-4-ggml-model-f32.bin'
...
[INFO] stable-diffusion.cpp:3375 - start sampling
[INFO] stable-diffusion.cpp:3067 - step 1 sampling completed, taking 12.25s
[INFO] stable-diffusion.cpp:3067 - step 2 sampling completed, taking 12.22s
[INFO] stable-diffusion.cpp:3067 - step 3 sampling completed, taking 12.56s
...
sampling completed, taking 246.40s
Is that expected performance?(EDIT: Don't have open Blas installed, so that flag is no-op)
Big if true (and if the opposite, of incrementally removing it also works). There have been similar platform efforts in past, such as https://news.ycombinator.com/item?id=20985429 . For that one, the "massive ask to give up every programming language and database they’ve ever used to depend on a startups closed source platform" seems like the biggest hindrance to adoption.
He's also American, so moving between US and UK must feel pretty fluid.
Here FAANG it ranges from 18 to 24 weeks for women, 12 to 18 weeks for "non birthing parent" (men, or adopting parents).
There's State-by-state paid family leaves laws, which makes it usually around 12 weeks (men or women).
Read somewhere that listening to music can trigger MD, is that the case for you? (Do you get less MD if you don't listen to music that day?)
Curious what this mean/what this actually looks like
3 days ago one of the obsidian makers posted this: https://twitter.com/kepano/status/1675626836821409792
Key quote:
"These days I write using an app I help make called Obsidian, but it’s a delusion to think it will last forever. The app will eventually become obsolete. It’s the plain text files I create that are designed to last."
Seems to align with your philosophy
You have to raise everyone's wages, prices, etc until there's an equilibrium (finally too expensive for digital nomads? no more rising rent prices). But it's not something that can be done easily/overnight.
This is how these places would have "a lot to gain". If it's just rent prices going up, the nomads & the local landlords are the only one benefiting and everyone else is upset.
> If child free, you’re only vested in whatever it takes to get you to the grave comfortably
Aren't you saying the same thing as parent post? When someone's 60 to 80 years old, which generation is going to "to get them to the grave comfortably"?
Basic necessities like food, etc, no matter the current amount of automation to multiply outputs, you still need a baseline of working people.
https://courses.fetalmedicine.com/images/Course/Course.007-0...
Notice the y-axis is a log scale.
This may be offset if women froze their eggs beforehand, but, obviously not everyone can afford it, and also until recently, it wasn't really needed as couples would have children earlier in life.
That makes it too easy for API consumers to ignore. (Compared to human eyeballs browsing a webpage)
But you could make it part of the ToS: if you're using the API to present posts to users, you MUST show all the posts in the list, including the is_sponsored=true ones.
Competent product/sales person, and engineering person seems missing for this to succeed. Especially if they used to rely on the eng infrastructure of giants.
They seem smart folks so probably can cover the software engineering gap, especially with $100m to outsource some of the hard problems to some cloud infra.
But I'd be more confident if they had started with a co-founder that's more product/sales oriented.
> As of today, the total amount to be repaid is set at
> €21,000 for students who have chosen to join a government body for Year 4
> €31,000 for students who have not chosen to join a government body for Year 4
Such amount is not as insignificant in a country with high taxes and where peers in university have like $100/year tuition, it's not super high either (maybe 4-6 months of after tax salary for an entry level position)
But I didn't bother posting is as i think it's not relevant. IIT could set that amount at 1 rupee, or 1 lakh, or 1 crore. There's a threshold where this switches from "that's fair you're repaying tuition" to "oh now you're adding a steep penalty"
It's complex to set the right value: too high and you're forcing your graduates to choose a high paying job or a government job, but you're killing their chances of startup creation. Too low and IIT is leaving money on the table.
IIT have prestige. If they care about leveraging that prestige and incentivize their graduates to some lifestyle choices there's ways to do it.
One example is the elite French "école polytechnique" (think the #1 IIT equivalent in engineering): tuition is waived, you even get paid a stipend as a student... But if you don't serve in French public service for 10 years after graduation, you have to repay it.
https://programmes.polytechnique.edu/en/ingenieur-polytechni...
> There are no study costs for French students, on the condition that they work for the civil service for at least 10 years within the 20 years following their graduation. Otherwise, they must repay the tuition fees
That would work as a "brain drain" tax.
Agree 100% here, with the assumption that the system is healthy to start with. And it's safe to assume it to be true for startups, which is the focus of PG's tweet. (If it's not, that start-up is already dead or close to be)
For startups, a competitive advantage is to do things that don't scale while you can. This includes "O(n²)" non-scalable communication, since n is still very small. And that kind of communication obviously works better in-person.
Late stage companies where async communication and slow-to-move things is the norm (due to the switch to a scalable communication method) actually benefit from remote work if the in-person office was kind of unhealthy for some people. Unhealthiness can be from the unavoidable slowness of communication/bureaucracy in a 50k+ employees company, or it can be from a rotten local culture in the office closest to you.
For this type of companies, remote options makes sense, the same way that expanding to other regions made sense. At this stage your communication is by nature scaled up, slower, and was already spanned across multiple buildings in original campus, so why not remote?
Literally today, most schools start at the exact same "time" (8:00am). But there's a big shift in actual solar time if you're on the far east or far west of a given time zone: https://i.dailymail.co.uk/i/pix/2014/03/03/article-2572317-1...
Do we see across the US a huge SAT score difference between kids on the most east of Pacific time and kids on west of Central time, and between kids east of Central time and west of Eastern time?
And within a the states that are pretty uniform (not the costal ones, as the socioeconomic differences would probably be too big), is there a big difference of SAT scores between the eastern-most highschoolers and Western-most highschoolers?
Per https://twitter.com/emollick/status/1194301106584616960:
> People in groups self-censor and anchor on each others ideas, reducing creativity.
Also can be anti-inclusive when "anchor on each other idea" is a euphemism for that guy repeating what a more shy person said, but louder, stealing credit.
With remote, usually comes a more doc-driven culture, where someone's original idea can be easier to identify and make sure they get credit.
The main difference is that AI extension, by design, send the content of the pages you browse to a server.
A malicious "calculator" extension could also send all the content to a server, and extension users don't really have an idea of what each extension is actually doing.
So skip the "Malware posing as AI browser extension" section, it's same kind of security issues as a malware calculator extension.
The legitimate AI extension's problems are more interesting.
Article wastes a bit more time on other security issues you get from using AI LLM in general. Those apply whether you're using a browser extension or chat.openai.com directly.
The valid point that applies to narrowly AI browser extension are:
1) it could send sensitive data you wouldn't have sent otherwise. Most people would know what they're doing when they explicitly paste the stuff on chat.openai.com. But when it's now automated via the extension DOM scraping, it's a bit harder to realize how much you're giving away.
2) And the hidden text prompt injection. That's interesting as now your attacker could be the website you browse, if you have configured too many plugins (Zapier plugin giving access to your email)
These 2 parts of TFA are imo novel security issues that only exist with AI browser extension, and are interesting.
Another issue I've had in newest file explorer, (probably a bug, unless they're trying to get rid of ".txt" files), is this one also reported on Twitter by "MittringMartin":
https://twitter.com/MittringMartin/status/166378202005572812...
can't do "right click" > "New" > "Text document".
Happens on my system too, a stock windows install on a surface laptop, for which I've never messed around with any obscure settings.
It's also gone from the "shift right click" menu (aka the good old menu)...
What's more common in practice ? Widely used libraries introducing critical vulnerabilities in 2023. Or old vulnerabilities being discovered/exploited, and patched?
> unless you as the upstream app developer decide not to build against the latest security updates of your dependencies; but if that's the case, that's a completely separate issue from dynamic vs. static linkage.
That's precisely the problem isn't it? What if "You the upstream app developer" is on a long vacation, or abandoned the project? Multiply this by N where N is the number of different app developers.
Compare with "you as the OS admin". If you're using your system, you update that vulnerable dynamic library, and you're done.
If users of other machines are on vacation or abandoned their machine, that's not your problem. Your system is secure.
So I feel parent's point had everything to do with dynamic vs. static linkage.
> a software glitch was responsible for the incident. According to the Seattle Times, the tailstrikes occurred largely as the result of a bug in a program sold by a Swedish firm called DynamicSource. The program is supposed to deliver “crucial weight and balance data” that pilots enter into their flight computers to help determine stuff like “how much thrust the engines will provide and at what speed the jet will be ready to lift off.” In this case, the tool delivered faulty data that “seriously undervalued [the] weights for the airplanes.”