gield

For people with their iDevice set to the Swiss/Austrian German or Belgian Dutch locales, Apple Translate initially didn't even offer their languages in Apple Translate (i.e. not even German or Dutch). Only after internal complaining did they allow Swiss/Austrian German users to use German in Apple Translate.

Cervélo is not an Italian company, they're Canadian. Its name is partially Italian as it's a portmanteau of cervello (brain in Italian) and vélo (bike in French).

That is the paper Krapivin read in ~2023 and inspired him. The actual paper with the breakthrough is from January 2025: https://arxiv.org/abs/2501.02305.

About 55% of Belgians speak Dutch as their native language. 45% French and <1% German.

Yes, that's explicitly mentioned in the blog post:

>In s1, when the LLM tries to stop thinking with "</think>", they force it to keep going by replacing it with "Wait".

It's not that easy to just send him a laptop [1]:

>>I'd caution against posting a laptop to Bangladesh as there is a very real chance it'd spend the next six months sat in customs while an official looks for a pay off.

>Yeah, this is the reason I wasn't too keen on having stuff shipped from outside.

[1] https://github.com/OXY2DEV/markview.nvim/issues/218#issuecom...

You wouldn't feel the stiffness of an aluminium frame as much on a track. On the road, stiffness is a more important factor for rider comfort.

Waze Beacons [1] do this for car tunnels. Some underground systems, e.g. in Barcelona IIRC, also use BLE beacons to help phones position themselves.

[1] https://support.google.com/waze/partners/answer/9416071

Outside of North America, it would be bizarre to not include VAT. Anytime you see the € or £ symbol, you can assume it's VAT included.

The article explains it very well.

Indeed, I worked at Apple in Spain where the offices and stores are part of general unions. In my office, there were 2 unions present.

If you want a deep-dive into the mathematics as well, I really liked Ray Tracing from the Ground Up by Kevin Suffern [1]. The book focuses on the theory and mathematics of ray tracing, and uses C++ to build a ray tracer. The code however is simple and explained really well so you could adapt it to the programming language of your choice.

[1] https://www.oreilly.com/library/view/ray-tracing-from/978149...

Someone will definitely see it, but maybe not take action on it. Now that it's on HN, it will definitely get more traction.

FYI, your links says "Image not found"

= 1 gram of carbs per minute of training

The EU fined Meta for 1.2 billion dollars last year, the largest GDPR fine ever [1]. And last month, the EU announced another investigation into Meta [2].

[1] https://www.edpb.europa.eu/news/news/2023/12-billion-euro-fi...

[2] https://ec.europa.eu/commission/presscorner/detail/en/ip_24_...

Non-Spaniards always overestimate the importance of tourism in Barcelona.

3.9% of Barcelona's GDP in 2022.

Source: https://www.statista.com/statistics/1346730/tourism-contribu...

5.4% of Catalonia's GDP in 2022.

Source: https://economia.gencat.cat/web/.content/70_economia_catalan...

11.6% of Spain's GDP in 2022.

Source: https://www.bde.es/f/webbe/GAP/Secciones/SalaPrensa/Interven...

>solely Texas is larger than ENTIRE Western Europe

Definitely not, Texas is ~700,000 km2 while Spain is ~500,000 km2.

If you ever go to Barcelona, you will see slogans like "tourists go home, immigrants welcome" written everywhere.

>It’s accepted that the US movies and television industry can’t get contemporary Germany right

The US movies and television industry can rarely get any language right. This reminds me of the Dutch-speaking scene in Oppenheimer which no Dutch speaker can understand [1]. It's like a sped-up pronunciation of a nonsense Dutch sentence by an English-speaker trying to speak a horrible German accent.

[1] https://www.youtube.com/watch?v=lFORgWaYrBU

>I'm hoping that, one day, we get a tutorial for VoiceOver

Maybe it's not feasible for you but if you're ever near an Apple Store, you could definitely call them and ask whether they have an accessibility expert you could talk to. In the Barcelona Apple Store for example, there is a blind employee who is an expert at using Apple's accessibility features on all their devices. He loves explaining his tips and tricks to anyone who needs to.

>A good old markov chain can simulate the average reddit thread

Someone did this on Reddit 9 years ago. It was remarkably good.

https://old.reddit.com/r/SubredditSimulator/comments/3g9ioz/

I simply asked whether it was GPT-4.5 or GPT-5:

>Are you GPT-4.5 or GPT-5?

>I'm based on GPT-4. There isn't a "GPT-4.5" or "GPT-5" version specific to my model. If you have any questions or need assistance, feel free to ask! I'm here to help with information up to my last update in November 2023. What would you like to know or discuss?

This html5 version has also been around for longer than a decade already. It's what inspired me to take a class on Genetic Algorithms and Evolutionary Computing in university back then.

OCR has been around for a few decades by now: https://nl.wikipedia.org/wiki/Optical_character_recognition

Sorry OP, you seem to have a wrong understanding of what an LLM is, does and how it's trained. And I know that an LLM is not a chatbot, no worries.

At most, your project is a simple language model, but definitely not a large language model.

After looking at your code, you also seem to have a wrong understanding of embeddings. Other people in this thread have already shared great resources that offer good explanations on these topics.

You seem to be very dismissive of Markov chains (or HMMs) while they've been used in NLP for years and produce really great results. Your project seems to use some concepts of Markov chains but more simplified. Here [1] is a random article that does next-token prediction using Markov chains. It's basically what your project does but in a more scalable and probabilistic way.

>Use cases include: Auto-completion, auto-correct, spell checking, search/lookup, conversation simulation (chatbot), and more.

These use cases are not possible using the concepts used in your project.

[1] https://bespoyasov.me/blog/text-generation-with-markov-chain...

Indeed, the last paragraph seems totally made up. The whole panel was stolen, not half of it. And Goedertier never demanded any ransom, the thief did but the article assumes he's definitely the thief. It's still unclear whether he was involved, he was a suspect because during the final minutes of his life, he mumbled some words about being the only person who knows where the stolen panel is.

The paragraph in question, for posterity:

>The artwork was stolen yet again in 1934 by stockbroker and businessman Arsène Goedertie. He sent a group of men to the cathedral one evening to steal the bottom left panel known as The Just Judges. Goedertie had sliced the panel in half, leaving the other half for the police to find, and demanded a ransom for the panel. The Belgian Minister refused, so the panel is still missing to this day. You can still check out the entire 12-panel Ghent Altarpiece on display, with an impressive replica of the missing panel.

AI is a lot older than Reddit.

2000 was a leap year so it would not fail in 2100. But it would indeed fail in 2200/2300/2400 as 2100/2200/2300 are not leap years.