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

OpenEuroLLM

openeurollm.eu

Post reply on HN
Print view View original

101–110 of 181 posts

Re: OpenEuroLLM

#101

Post by belter »

sdsd wrote: post #63
belter wrote: post #36

Earlier quoted context omitted.

Safety and no School shootings, Universal Health Care, Worker Rights, almost free Universities. The USA in the 23th century...

i see this kind of https://en.wikipedia.org/wiki/Whataboutism a lot. i think that you can imitate good things about US innovation without having lots of school shootings.

I’m pointing out that overall compensation isn’t just the salary. In Europe, benefits like universal healthcare, free universities, and strong worker rights add significant value that balances lower wages.

Re: OpenEuroLLM

#102

Post by amarcheschi »

olejorgenb wrote: post #98
mrshu wrote: post #69

It is worth noting there is _another_, completely unrelated project (also) called *EuroLLM* that is also EU funded which not only shares many of the same goals, but has already fulfilled many of them: 1. large multilingual dataset 2. open science approach 3. competitive performance Here is the HF blogpost that introduced it in December 2024 (along with various benchmarks): https://huggingface.co/blog/eurollm-team/eur…

Homepage: https://sites.google.com/view/eurollm Deliverables: - A series of models of different sizes for optimal effectiveness and efficiency (1B, 9B and 22B) trained on 4T tokens - A multimodal model which can process and understand speech or text input - Full project codebase available to the public with detailed data and model descriptions I can't find the codebase yet though

Results don't seem that bad for 9b https://huggingface.co/blog/eurollm-team/eurollm-9b

Re: OpenEuroLLM

#103

Post by sunshine-o »

rafram wrote: post #30
sameermanek wrote: post #20

They also had some search engine announced with similar name like "openeurosearch" or something close to that. That project too seems dormant lately. They just announce things and then the train leaves the station.

That was just an Ecosia rebrand, not an official EU thing: https://betterweb.qwant.com/en/2024/11/08/ecosia-and-qwant-j...

The official EU "Google alternative" was Quaero (not Qwant which is something else). Announced in 2005 and ended in 2013.

I don't believe anything was ever made public.

[0] https://en.wikipedia.org/wiki/Quaero

Re: OpenEuroLLM

#104

Post by varjag »

rafram wrote: post #73
coalbin wrote: post #42

Earlier quoted context omitted.

Maybe you can't download their weights, but you can literally try out their products right from their homepage. What's your point?

OK, if you prefer: https://web.archive.org/web/20151211215507/https://openai.co... It's normal to announce things long before they're actually available to end users. This is not some unique evil of the EU bureaucracy - if anything, it's very corporate of them.

You just don't get it. When it's announced by the EU, it's red tape PDF crunching soulless bureaucracy wasting our time. When it's said by a glorified CPA of an American unicorn startup it's the vision, innovation and triumph of entrepreneurial spirit.

Re: OpenEuroLLM

#105

Post by rhubarbtree »

dagenleg wrote: post #100
ta12653421 wrote: post #75

Earlier quoted context omitted.

for 1.87m per project, you get in EU rather 15 - 20 people :) (salaries are low here)

At least in France, where they have PhDs which last only 3 years, a years of PhD would cost ~45K EUR in gross salary (granted the student gets around half of that after tax), then let's say ~10K travel and consumables costs, then add up the inevitable 20% overhead costs and now you're looking at around 200K for the shortest possible frugal 3 year PhD.

At least in the UK, overheads are usually over 100%.

Re: OpenEuroLLM

#106

Post by cess11 »

Legend2440 wrote: post #74
rafram wrote: post #28

Earlier quoted context omitted.

> Press release https://openai.com/news/company-announcements/ > a consortium of 20 research institutions https://aimagazine.com/machine-learning/google-invests-in-ai... > awarded the STEP (Strategic Technologies for Europe Platform) seal https://openai.com/index/strengthening-americas-ai-leadershi... > Lots of grandiose self-congratulations https://x.com/sama/status/1891533802779910471 > All with nothing to run, dow…

OpenAI has a real, groundbreaking product. This has... a statement of intent to try to copy that product. Not remotely the same.

This is about industrial research, not about some product.

Members of the project have previously produced both niche and general models, but without the arrogance and bluster of usian corporate subcultures.

Re: OpenEuroLLM

#107

Post by hanshansen43 »

sunshine-o wrote: post #88
dkyc wrote: post #3

But can I run it on Gaia-X? This really reads like a parody. Press release, “a consortium of 20 research institutions”, “awarded the STEP (Strategic Technologies for Europe Platform) seal”. Lots of grandiose self-congratulations. All with nothing to run, download or try of course.

What I am gonna say here is not a political point but I hope someone can point me the pattern (and some something to read about it) I have observed with for example the EU. Yes it sounds like a parody or an onion piece. We know the European search engine, cloud, blockchain never got anywhere. I don't even believe that anybody ever really tried. Now you have to put yourself in their head for 2 minutes and here is what…

As a European, the sad reality is that I see parallels with the late-stage Soviet Union and its satellite states.

NOT when it comes to the level of violence and repression or quality of living. Those two things are world-class.

But in the sense that there's a more or less unelected political establishment that's

a) Recursive: It does things only to show them off to itself.

b) Not exposed to real-world consequences.

c) Has a non-falsifiable pretense to validate whatever they do and caution against undoing whatever it is. For the soviets, it was anti-capitalism. For the EU it's some notion of safety or sustainability.

d) Inadvertently benefits itself and other elites and harms the people they pretend to protect.

My hope is that as a democratic institution, the EU is capable of reform.

Re: OpenEuroLLM

#108

Post by bee_rider »

huijzer wrote: post #71
seydor wrote: post #41

> OpenEuroLLM has a total budget of €37.4 million of which €20.6 million comes from the Digital Europe Programme. So on average 1.87 million per participating institution which might amount to funding ~5 PhD students per institution. Not bad for a training program. The project has been awarded the Sovereignity Seal, an EU mark of Excellence before it even started. This is truly in accordance with european values, whe…

> This is truly in accordance with european values, where we reward participation and proclamation. Yes. In my experience the government is happy with "looks good doesn't work" as long as it truly looks good.

Thank goodness we have startup culture in the US, instead we can have “looks bad, doesn’t work, but Microsoft bought it so…”

Re: OpenEuroLLM

#109

Post by belter »

amarcheschi wrote: post #102
olejorgenb wrote: post #98

Earlier quoted context omitted.

Homepage: https://sites.google.com/view/eurollm Deliverables: - A series of models of different sizes for optimal effectiveness and efficiency (1B, 9B and 22B) trained on 4T tokens - A multimodal model which can process and understand speech or text input - Full project codebase available to the public with detailed data and model descriptions I can't find the codebase yet though

Results don't seem that bad for 9b https://huggingface.co/blog/eurollm-team/eurollm-9b

"...EuroLLM-9B was trained on approximately 4 trillion tokens, using 400 Nvidia H100 GPUs on the MareNostrum5 supercomputer..."

Re: OpenEuroLLM

#110

Post by simion314 »

ffsm8 wrote: post #70
simion314 wrote: post #43

Earlier quoted context omitted.

So CERN and similar? you think it could have been done without any coordonation? Like 2 guys faking under they make it USA style ?

Well, what actual value did cern produce beyond theoretical research with no application in sight?

And NASA? they just send robots in space to make pictures? Not sure why you are here and not on TikTok with such anti science ideology.
Post reply on HN