Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

EuroLLM: LLM made in Europe built to support all 24 official EU languages

eurollm.io

Post reply on HN
Print view View original

431–440 of 629 posts

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#431

Post by unscaled »

td540 wrote: post #93
ChrisMarshallNY wrote: post #84

Earlier quoted context omitted.

Flemish? I remember watching a TV show in Flemish ( Hotel Beau Séjour [0]), so it's prevalent enough to invest that kind of money into. What about Basque? Is that too controversial? [0] https://en.wikipedia.org/wiki/Hotel_Beau_Séjour

like British English vs US English, Flemish is a dialect of dutch

I think sentence should be easily readable to Flemish speaker: "A shprakh iz a dialekt mit an armey un flot"

https://en.wikipedia.org/wiki/A_language_is_a_dialect_with_a...

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#432

Post by panick21_ »

InsideOutSanta wrote: post #205
nonethewiser wrote: post #25

Earlier quoted context omitted.

"Why" is a fair question but are you surprised? Europe is consistently behind in tech. Europe has about 1.3 times the population of the USA and about 75% of the GDP yet EU tech output is a very small percentage of US tech output. We are not talking about 70, 50, 30, or even 20%. It's a drop in the bucket. >The seven largest U.S. tech companies, Alphabet (Google), Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla, are…

> The seven largest U.S. tech companies (...) are 20 times bigger than Europe’s seven largest, and generate 10 times more revenue. I'm going to guess that this part is intentional. Europe tends to be more aggressive in enforcing antitrust laws. Economically, Europe's goal isn't to have the biggest companies but to have more smaller companies. So you're not going to get companies like Google, but you will get companie…

Antitrust laws are not the reasons for more smaller companies. Getting an antitrust case off in Europe is very hard, not US hard but still hard. The reasons are more complex then that.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#433

Post by unscaled »

punnerud wrote: post #121
adzm wrote: post #2

For those curious, the 24 official languages are Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Irish, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Slovak, Slovenian, Spanish, and Swedish. Maltese, interestingly, is the only Afro-Asiatic derived language. Hungarian, Finnish, and Estonian are the three Uralic languages. All the others ar…

Norwegian is also included, based on the model card: https://huggingface.co/utter-project/EuroLLM-9B

Considering there are two different official written forms of Norwegian, that's not really saying enough, but I guess they mean Bokmål.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#434

Post by unscaled »

charlieyu1 wrote: post #364
loandbehold wrote: post #10

Aren't all frontier models already able to use all these languages? Support for specific languages doesn't need to be built in, LLMs support all languages because they are trained on multilingual data.

Training is a very different thing. Can’t speak for European, but LLMs are often much worse in Japanese because tokenisation used Unicode and a single Japanese character often has to be represented by more than one token

I think you meant to say that tokenization is usually done with UTF-8 and a single Japanese character generally takes 3 or more code units (i.e. bytes). Unicode itself is not the culprit (in fact, even with UTF-16 tokenization, most Japanese characters would fit in a single code unit, and the ones that won't are exceedingly rare).

I have to admit I have not encountered significant mistokenization issues in Japanese, but I'm not using it on a daily basis LLMs. I'm somewhat dobutful this can be a major issue, since frontier LLMs are absolutely in love with Emoji, and Emoji requires at least 4 UTF-8 bytes, while most Japanese characters are happy with just 3 bytes.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#435

Post by megous »

Byamarro wrote: post #294
tensor wrote: post #109

Earlier quoted context omitted.

No, that's not how training works. It's not just about having an example in a given language, but also how many examples and the ratio of examples compared to other languages. English hugely eclipses any other language on most US models and that's why performance on other languages is subpar compared to performance on english.

There's actually a research showing that llms are more accurate when questions are in Polish: https://arxiv.org/pdf/2503.01996

My first impulse is to say that some languages have better SNR on the internet. (less garbage autogenerated or SEO content compared to useful information)

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#436

Post by biztos »

hebelehubele wrote: post #157
outside1234 wrote: post #149

Earlier quoted context omitted.

This raises an interesting question. Is there only one dialect of German in the LLM? My understanding is that the German German and Austrian German dialects are significantly different.

My German teacher always claimed that Swiss German and German German (Hochdeutsch) were so different that she needed subtitles to understand it, and she didn't understand why they weren't considered separate languages.

Even Swabian, a dialect spoken mostly in Germany, is almost unintelligible to non-native speakers when spoken by the natives of a certain age.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#437

Post by hopelite »

mikrl wrote: post #81
Vinnl wrote: post #35

Earlier quoted context omitted.

Tomorrow there are elections in the Netherlands, and two parties are proposing adding Frysian to that list: https://neerlandistiek.nl/2025/10/kies-voor-taal/ Best get to retraining those models.

As a Brit I feel very at home when hearing/reading Dutch and Frisian. It’s a reminder that England and the Low Countries share a lot of close history all the way back to Anglo-Saxon times; of being fishers, traders, burghers and mercenaries moving around the North Sea chasing opportunities, spreading and augmenting languages. “Brea, bûter en griene tsiis is goed Ingelsk en goed Frysk”

That’s because all those languages are all essentially rooted in the languages/dialects of the Germanic tribes. It is why the Dutch get their English name from the German for German, Deutsch; and Nederland (Neder = Low) is German/Dutch for the Lowland Deutsch.

I’m sure everyone is aware that English comes from Anglish, i.e., the Angles as in the Germanic tribe.

Deutsch is derived from proto-germanic (as best we can tell) þiudiskaz, meaning “the people” i.e., the group of the different self associating tribes. It gets far more interesting in that it seems many of the strong dialects of especially southern Germany, Austria, and England have in fact retained some very old words and pronunciations that were lost in more standardized, conformed, and perverted dialects.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#438

Post by o11c »

whimsicalism wrote: post #170
sofixa wrote: post #155

Earlier quoted context omitted.

> but I started my company in the US because everything is so much easier here Which part is easier? That you have 50 different states with slightly varying laws to consider (e.g. Californian Data protection)? That you have a byzantine system of "benefits" to choose and manage? And compared to where? Germany or Estonia or Sweden or Spain? The complexities will vary wildly depending on the country (kind of like in the…

"That you have 50 different states with slightly varying laws to consider (e.g. Californian Data protection)?" there are certain sentences you can just tell would never be written by an American lol

Americans are definitely aware of it in some areas, e.g. any sort of licensing.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#439

Post by bbno4 »

dzikimarian wrote: post #130
whimsicalism wrote: post #57

[flagged]

While grant process in EU isn't fun, I think Levels has bit of an ego issues. He mentioned that if he had issues like that on eg X, he would see Elon himself in the replies. While he is great at converting his influencer status to income in his micro-SaaS projects, I don't think running ad-fueled browser games on state-sponsored super computer should be really aim of these grant programs.

Fully agree with you here.

Re: EuroLLM: LLM made in Europe built to support all 24 official EU languages

#440

Post by veqq »

kpil wrote: post #405
RobotToaster wrote: post #97

Earlier quoted context omitted.

If you've ever read anything written in old English, it's a even closer to Dutch.

Old English looks more or less like old Norse to me. Or old Scandinavian as we say in Sweden...

Old English and Old Norse are mutually intelligible (especially after you realize the precise correspondences like un- = o-). Gunnlaugs Saga explicitly says the English and Norse are of one tongue and features a Norse poet singing to an English king. As another example, Ohthere of Hålogaland (Norway) visited King Alfred's 9th century English court and simply spoke to them in his own language:

https://web.archive.org/web/20170530232902/https://blogs.bl....

> Whoever preserved this story was also curious about Ohthere’s descriptions of where the Angles had lived ‘before they came into this land’ (England). Members of Alfred's court remembered that their ancestors came from mainland Europe, and they wanted to learn more about the lands which they identified as their own places of origin.

The scribe explicitly wrote things like "he said krán which we call crein" showing they were speaking in their own languages. It's even clearer if you consider our standard Old English is West Saxon from 850 and our standard Old Norse is from 1250 in Iceland (more different than the Danish variety of most Scandinavians in England). At the same time point,they would have more similarities (8th century Danish had wír before w turned to v).

https://en.wikipedia.org/wiki/Ohthere_of_H%C3%A5logaland

Post reply on HN