The misleading prompt cases are inadvertently testing the model's ability to filter out noise from its instructions. That could be a benchmark on its own. The correct response is to flag the inconsistency and ask for clarification.
HN user
zvolsky
tmux resident, developer, FOSS contributor.
Occasional blogger at zvolsky.org.
hn@zvolsky.org
Hey, Grant here - I wrote most of the 2.0 migration code together with Aeneas. Thank you for the writeup team Cloudflare!
After investigation, we discovered that there was an issue in one of the Hydra migrations that corrupted the state of certain valid OAuth sessions, which resulted in the migration marking them as invalid.
Was this one of the open source migration files? While I'm no longer involved in the project, I'd be curious to know if it's been addressed upstream.
The idea of HN being dismissive of impactful technology is as old as HN. And indeed, the crowd often appears stuck in the past with hindsight. That said, HN discussions aren't homogeneous, and as demonstrated by Karpathy in his recent blogpost "Auto-grading decade-old Hacker News", at least some commenters have impressive foresight: https://karpathy.bearblog.dev/auto-grade-hn/
In May this year, the 2x48GB Crucial SODIMM kit sold for £180. Today, the same kit is £275 in the Crucial Amazon store, and sales are limited to one unit per customer. The free market price seems to be over £500. Not a good time to be building a Mini PC or DIY laptop.
Impressive feat by Waymo, though it could have been 100-200ms faster had it noticed not only the change in trajectory, but also the jerky motion that destabilized the scooter.
For a truly superhuman performance, it might even register the brick and anticipate the scooter's accident.
Does this mean that Minimum TDP = maximum power consumption* at the CPU's lowest power configuration?
* Averaged over some time window relevant for cooling
Hey, thanks again for the talk and for answering my fork bomb question with a live demo!
The word `official` was added by OP to hint at the fact that zx2c4 is the kernel contributor who created WireGuard.
It isn't clear whether it was the milking robot or the farmer's computer that was compromised.
A more measured take on the relationship between science and religion: https://philosophy.stackexchange.com/a/29763
Haha, I've always enjoyed being at the end getting less attention from teachers. If the data merely shows a correlation, it may as well be explained by us at the end being under less pressure.
Most useful is relative. The book dealers who gave me my first ever contract for an automated leaflet generator that generated customized leaflets for each dealer would consider that the most useful. It's been over a decade and they still use it every day. Some other projects that come to my mind:
- an open source OIDC authorization and identity server (Ory Hydra and Ory Kratos)
- a system that uses computer vision to track food waste in industrial kitchens, helping them understand and minimize food waste
- a live milk quality estimator for a milking robot
- a browser extension called memorize. I wrote this reusing code from another contract I did while still in secondary school, and I'm mentioning it thanks to a user named lush berry who wrote "literally the most helpful thing i ever found on the internet. it helps me memorize stuff even when i'm procrastinating which is amazing. however, i have a lot of suggestions, does anyone know how i can contact the makers of this extension?". I wish I had the time to listen to these users and make the changes they want.
Thank you for the explanation!
Because my card was issued by MasterCard, not Amazon. The technology between the card reader and my card account is determined by MasterCard, and in theory MasterCard don't need to give Amazon any identifying information to process a payment.
One of my relatives, uncle George, was fired ~30 years ago. His most valuable possession was a gold watch that baroness Coudenhove gave to a her maid 2-3 generations earlier. I don't remember who exactly the maid was, but she was somehow related to the family. In order to raise funds to buy his own tipper truck, George went to the Charles Bridge in Prague and found a buyer for the watch, a German tourist. The money was at least 50% of the price of the truck. These were the first steps on a long journey leading to a successful business.
This was in a country with high inflation that's just gone through a regime change. Did he get a good price for the watch? Yes, at that moment, the truck was more valuable than the watch.
Why does it have to give you this condescending spiel about "offensiveness" or some other subjective reason?
The response quoted above makes no mention of offensiveness, and I explained the need to decline certain requests above.
I understand your sentiment, and I would agree if it was the stock response we are used to from ChatGPT. Unlike the condescending stock response from ChatGPT, however, the response in question is to the point, honest, and provides useful feedback.
It is offensive if you take the output personally. You are interacting with the model, but the model isn't interacting with you. The model doesn't know who you are. It could be the bad actors currently confined to the spam folder of your email making these requests, and the model wouldn't know the difference.
Even if you are using Google Analytics, you can skip the consent banner by denying on behalf of the user. GA will then update your counters without doing anything that would require consent.
ad_storage='denied'
> Requests are sent through a different domain to avoid previously set third-party cookies from being sent in request headers.
> Google Analytics will not read or write Google Ads cookies, and Google signals features will not accumulate data for this traffic.
> IP addresses are used to derive IP country, but are never logged by our Google Ads and Floodlight systems and are immediately deleted upon collection. Note: Google Analytics collects IP addresses as part of normal internet communications.
https://support.google.com/analytics/answer/9976101?hl=enYour frustration is understandable, but there is no fault on either side. Language evolves and different fields borrow and evolve the meanings of words all the time. Not even the `neuron` that you're thinking of is the original meaning of the word.
w3schools used to be one of the best resources ~15 years ago. Today I intuitively avoid it because
1. The ads make the fan on my laptop spin (I prefer to avoid ad-heavy websites over using adblock)
2. The 'Try it yourself' button is fake. It doesn't actually run the code and doesn't let you edit it.
3. It is not a good reference. To give an example, the last time I visited w3schools was upon searching "react ul li", which landed me on the page https://www.w3schools.com/react/react_lists.asp . Ironically, the reason why I was searching the term, was to find advice on setting the "key" prop in on list items. The w3schools example doesn't even mention the prop and produces errors. Compare that to the official reference which ranks much lower in Google search: https://react.dev/learn/rendering-lists#where-to-get-your-ke... .
I've used Xero, QuickBooks, and Freeagent for microentity UK LTD accounts.
Xero is the champion of dark patterns and gaslighting when it comes to billing. The listed price fails to take into account their notice period and they won't let you export your data without resubscribing (with a notice period for the new subscription!).
I tried QuickBooks just briefly and didn't adopt it because I felt like the interface is hiding important details and doing things in the background that I want to be explicit and possibly done differently IIRC it wouldn't let me create an invoice without GBP conversion which isn't required for non-VAT invoices. This led me to creating my own invoice editor that I'm happy with.
FreeAgent has everything I need - bank feeds, streamlines the whole end of year process submitting everything to HMRC. It doesn't implant a vendor lock-in abstraction layer between the user and the accounts. Xero and QuickBooks feel like they are designed to be used exclusively by full-time accountants who specialize in their software package, whereas FreeAgent can be used by a founder educated in accounting who prefers to do the accounting on their own.
It is possible that this tragic accident would have happened even if the operator paid full attention, but that is besides the merit of this case. Their job was to supervise an autonomous driving test, and they did not pay full attention at the critical moment, with no justifiable reason for this lack of attention:
"Records indicate that streaming began at 9:16 pm and ended at 9:59 pm. Based on an examination of the video captured by the driver-facing camera, Vasquez was looking down toward her right knee 166 times for a total of 6 minutes, 47 seconds during the 21 minutes, 48 seconds preceding the crash. Just prior to the crash, Vasquez was looking at her lap for 5.3 seconds; she looked up half a second before the impact." - https://en.wikipedia.org/wiki/Death_of_Elaine_Herzberg
When Chrome launched, its rendering speed was in a league of its own, delivering immediate results without the need to repeatedly click a link for faster loading. The design was clean and intuitive compared to alternatives such as Firebug and Fiddler the HTTP debugging proxy.
Our best computer networks teacher from secondary school decided to leave following complaints from parents, who believed that the tests were too difficult. As I remember, about 3/4 of our class had a tough time. Honestly, the tests weren't all that hard. I suspect there was some crowd effect taking place, which resulted in people collectively giving up. This turned out to be a significant blow to the standard of our education. The subsequent teachers were neither as knowledgeable, nor as effective in teaching the subjects.
Quoting my comment from another thread:
... this whole debate is reminiscent of Karel Čapek's War with the Newts [1936]. In particular the public discourse from a time before the newts took over. "It would certainly be an overstatement to say that nobody at that time ever spoke or wrote about anything but the talking newts. People also talked and wrote about other things such as the next war, the economic crisis, football, vitamins and fashion; but there was a lot written about the newts, and much of it was very ill-informed. This is why the outstanding scientist, Professor Vladimir Uher (University of Brno), wrote an article for the newspaper in which he pointed out that the putative ability of Andrias Scheuchzer to speak, which was really no more than the ability to repeat spoken words like a parrot, ..." Note the irony of the professor's attempt to improve an ill-informed debate by contributing his own piece of misinformation, equating newt speech to mere parrot-like mimicry.
While I remain undecided on the matter, this whole debate is reminiscent of Karel Čapek's War with the Newts [1936]. In particular the public discourse from a time before the newts took over. "It would certainly be an overstatement to say that nobody at that time ever spoke or wrote about anything but the talking newts. People also talked and wrote about other things such as the next war, the economic crisis, football, vitamins and fashion; but there was a lot written about the newts, and much of it was very ill-informed. This is why the outstanding scientist, Professor Vladimir Uher (University of Brno), wrote an article for the newspaper in which he pointed out that the putative ability of Andrias Scheuchzer to speak, which was really no more than the ability to repeat spoken words like a parrot, ..." Note the irony of the professor's attempt to improve an ill-informed debate by contributing his own piece of misinformation, equating newt speech to mere parrot-like mimicry.
Čapek, intriguingly, happens to be the person who first used the word robot, which was coined by his brother.
Would you elaborate on your comparison with Germany? OECD GDP, CPI inflation, and inflation forecast figures seem to contradict your argument.
The Elements of Statistical Learning. This book features some of the most interesting applications of mathematics. It was the likely reason why I started enjoying maths after many years of studying it.
The article talks about store shortages and becoming a poorer country. I haven't been affected by the shortages so I presume they aren't as significant as the media present them. As for getting poorer, it seems to be a Europe-wide problem - hardly related to Brexit.
As someone who has spent a significant amount of time in the UK and the Czech Republic (an EU country) since Brexit, both places are worse off now, but the Czech Republic is significantly more worse off than Britain. I have plenty of anecdotal evidence, but it suffices to look at inflation statistics, with Czechia at 15.1% and the UK at 9.07% for 2022. I haven't noticed any empty shelves, except for eggs, which were caused free-range chicken quarantine (according to a store worker). Maybe it's a coincidence, but as far as I can tell, this crisis has nothing to do with Brexit.