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

361–370 of 629 posts

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

#361

Post by lhoff »

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.

It depends. There is not one Swiss German but multiple subdialects. The language spoke around the Bern region very far away from German while the one from Zürich or Basel is much closer. Since there is no official written from they never really converged to a homogeneous language.

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

#362

Post by findyoucef »

arbuge wrote: post #119
Raed667 wrote: post #40

Earlier quoted context omitted.

Tunisians claim they can understand Maltese with minimum effort, is it reciprocal? How close is Maltese to arabic / tunisian dialect ?

Not sure which Tunisians are claiming this but they'd definitely need a lot more than minimum effort. Maltese split off from Arabic around 1k years ago. The two languages sound pretty different, and are written with different alphabets.

As an Algerian, I can confirm that Maltese is surprisingly easy to understand. I was genuinely shocked the first time I heard it because the similarities are so obvious. Many Arabic dialects are also written using the Latin alphabet, especially online and on social media, so the different writing systems aren’t really a barrier at all.

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

#363

Post by neoromantique »

emporas wrote: post #160
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…

Also, commercial software is consistently behind from open source. I only use open source LLMs for writing (Qwen 32b from Groq) and open source editor of course, Emacs. If some people can write better using commercial LLMs (and commercial editors), by all means, but they put themselves at a disadvantage. Next step for me, is to use something open source for translation, I use Claude for the moment, and open source fo…

What a weird comment.

llama-3.1-70b-versatile is pretty good at translating though

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

#364

Post by charlieyu1 »

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

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

#365

Post by vanviegen »

troupo wrote: post #346
pier25 wrote: post #338

Earlier quoted context omitted.

Depends on the country. Opening a company in Estonia is very cheap but in Spain the manager/CEO needs to be an "autónomo" (like a self-employed tax status). This costs thousands of Euros per year. Something like 2,400-30,000 Euros per year, every year, forever.

And that's probably one of the big obstacles in the EU: there's no common ground for these things. At least this will hopefully be addressed: https://www.reuters.com/business/eu-propose-uniform-rules-st...

What does it matter that the rules for establishing differ per country? I'm only founding in one of them.

The article is unclear, but is probably referring to making it easier for startups to offer products in other EU countries.

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

#366

Post by vanviegen »

whimsicalism wrote: post #152
deaux wrote: post #140

Earlier quoted context omitted.

> I think it is a complete misreading to point to protectionism as the reason for Chinese success, but having a big unified domestic market for consumers along with massive saving rates and capital controls probably does help. Capital controls are protectionist measures, but anyway, no. > Okay, what is the limiting factor? Let's look at which countries have a significant local software industry compared to population…

> Their market share is almost 1:1 linearly correlated to the degree of protectionism in those countries Seems like you actually believe this. I think our starting points on reality are different enough that we are not going to have a productive conversation, I wish you and other Europeans the best of luck in your protectionism-led growth strategy. Make sure to not discuss it with any pesky macroeconomists who might…

You seem to have accidentally left the actual content out of your comment.

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

#367

Post by carlosjobim »

sofixa wrote: post #168
whimsicalism wrote: post #104

Earlier quoted context omitted.

> When you talk to most EU business owners, even in tech, the limiting factor isn't regulations. This being the #1 reason is such a tired trope. Okay, what is the limiting factor? Because when I talk to EU business owners (admittedly, very few) - they point to lack of big EU capital markets, which is directly downstream of the policy environment. And when I talk to top EU human capital, they all point to the lack of…

> Okay, what is the limiting factor A few. A big part is that the EU is a collection of countries that (with very few exceptions) have different languages and laws. For a company to serve Spain and France, for instance, it would need to translate everything, hire local lawyers and customer support agents. Considering the much smaller size of the countries (biggest one is 70 million vs 330 million in the US), the oppo…

> Considering the much smaller size of the countries (biggest one is 70 million vs 330 million in the US), the opportunity for "unlimited" growth is limited.

If you manage to get 10 million customers, your business is already successful on a gigantic scale, and you should have all the know-how in taking on the world. The success of other people is rarely the reason why you are failing in your own life. Start somewhere, do something.

> The money being thrown around the US is like nowhere else on the planet.

That's true and it's awesome. In Europe money is only thrown to real estate owners and any enterprising people with a dream are cordially invited to fucking forget about it, shut up, and fall back in line. Even if they already have a proven track record. They take their idea to the United States and are treated incredibly well in comparison. Even if their business will only be a niche business with limited reach, like 99% of businesses.

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

#368

Post by redfloatplane »

Some cursory clicking about didn't reveal to me the actual corpus they used, only that it is several trillion tokens 'divided across the languages'. I'm curious mainly because Irish (among some other similarly endangered languages on the list) typically has any large corpus come from legal/governmental texts that are required to be translated. There must surely be only a relatively tiny amount of colloquial Irish in the corpus. It be interesting to see some evals in each language particularly with native speakers.

I think LLMs may be on the whole very positive for endangered languages such as Irish, but before it becomes positive I think there's an amount of danger to be navigated (see Scots Gaelic wikipedia drama for example)

In any case I think this is a great initiative.

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

#369

Post by vanviegen »

cbeach wrote: post #351
saubeidl wrote: post #296

Earlier quoted context omitted.

[flagged]

The EU Commission is appointed, not elected. Only the Parliament (MEPs) are elected. What’s worse, the parliament cannot originate law. Only the unelected Commission can do so. And they can do it behind closed doors. This is a setup that’s ripe for corruption.

It's true that they are appointed.

However, they're appointed by the EU Council (the heads of state, most of them elected, some appointed by a national parliament), and approved by the (elected) European Parliament.

At the cost of some transparency, this does make it possible to select a bit more for management skills instead of just campaigning skills.

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

#370

Post by simion314 »

imcritic wrote: post #215
dostick wrote: post #200

What good does it do by having only include formal languages? For example there’s no Russian, while there’s now at least 8 million ethnic Russians living in Europe.

[flagged]

Oh, typical Ruzzian victim comlex, their brain can't understand why all their neighbors and "brother slavs" hate them, brainwashing for generations made them think in unatural logic where you need to negate anything a Ruzzian says and then you increase the probability 100 times to be the truth.

Ruzzian = a Russian Zed patriot , we use this notation to acknowledge that there still exists a small percentage of educated Russians that are not Ruscists.

Post reply on HN