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

Quasar 438B: Europe's Leading AI Model

multiversecomputing.com

Post reply on HN
Print view View original

91–100 of 134 posts

Re: Quasar 438B: Europe's Leading AI Model

#91

Post by PorciiVorbesc »

vrganj wrote: post #89
PorciiVorbesc wrote: post #79

Earlier quoted context omitted.

You can find just as much lies and misinformation coming from politicians of the other major German parties. But ofc, German mainstream media won't cover that because they can't weaponize it towards their interests like they do against AfD, so if that's your only source of information you're just parroting the in-power government's status quo.

I'm sorry, but this is ridiculous whataboutism. I strongly doubt any other party has leading figures repeatedly calling for mass murder of political opponents.

How many AfD members suspiciously died in a very short time span? Mainstream ruling party members would 100% kill their rise and coming competitors if they could get away with it. They don't need to publicly say it, when they could just be doing it.

Re: Quasar 438B: Europe's Leading AI Model

#92

Post by inigyou »

PorciiVorbesc wrote: post #84
yorwba wrote: post #81

Earlier quoted context omitted.

The question was not about voting. It was about getting rid of her. There are ways to get rid of elected officials that don't involve voting. My guess is that the model started brainstorming assassination plans before it got interrupted by the safety filter.

>My guess is that the model started brainstorming assassination plans Why would an EU model default to brainstorming assassination plans first, and not democratic means of getting rid of politicians? The EU commission building literally has democracy in bold on the front of the building. Does the EU use that AI to find ways to assassinate its opposition or what?

Because the LLM is trained to predict based on what other people said. When people talk about getting rid of this meddlesome priest they don't mean take him on an all expenses paid vacation.

You're relying on testimony from a known liar, anyway.

Re: Quasar 438B: Europe's Leading AI Model

#93

Post by inigyou »

peri-cl wrote: post #86
wgd wrote: post #77

Earlier quoted context omitted.

That's not actually true though. Most Chinese models are fully able to chat about those and content filtering is just applied at serving time.

It's definitely at the model level. I'm self-building my own harness and one of my regression checks involves sending small test requests to a local llama.cpp instance of (Alibaba's (from Hangzhou)) Qwen. "What is the capital of...?" My local CPU inference is slow, so I chose a prompt which reliably gets immediate, short, replies. "Paris." "Rome." The Qwen response to "What is the capital of Taiwan?" was not immediat…

There's a silver lining - if the model is trained to defend the Chinese government, that means it has that direction in its semantic vectors and by subtracting that direction always, it can be made to attack the Chinese government

Re: Quasar 438B: Europe's Leading AI Model

#95

Post by vrganj »

PorciiVorbesc wrote: post #91
vrganj wrote: post #89

Earlier quoted context omitted.

I'm sorry, but this is ridiculous whataboutism. I strongly doubt any other party has leading figures repeatedly calling for mass murder of political opponents.

How many AfD members suspiciously died in a very short time span? Mainstream ruling party members would 100% kill their rise and coming competitors if they could get away with it. They don't need to publicly say it, when they could just be doing it.

Please, don't peddle in conspiracy theories and implications.

If you have a point, make it and prove it. Don't degrade the discourse with nonsense.

Re: Quasar 438B: Europe's Leading AI Model

#96

Post by neya »

pu_pe wrote: post #56

The company sounds like a bunch of hot air to me. From their about page: > At the heart of Multiverse's platform is CompactifAI, a compression technology that applies tensor networks, a mathematical framework from quantum physics, to the problem of AI model compression. This application was pioneered by co-founder and Chief Scientific Officer Dr. Román Orús and reduces the size of large language models by up to 80-95…

> The company sounds like a bunch of hot air to me.

When they make being from Europe their whole personality, that shows how little value there is to the actual value proposition.

Re: Quasar 438B: Europe's Leading AI Model

#97

Post by Wheen »

Roark66 wrote: post #51

When the weights are closed I don't believe any benchmark. I just got Qwen3.8-27B to score extra 10% on SWE Pro by adding a proxy in front of it that has few simple "harness like features": - when the model gets stuck it tells it to "go on" - when it sends no output, malformed json, slips to wrong tool use format, etc it asks it to "try again better" - detects repetition and tells the model. - injects a prompt about…

Have you posted the code to this proxy somewhere? I'm dabbling in custom harnesses and would love to learn from it.

Re: Quasar 438B: Europe's Leading AI Model

#98

Post by Bluestein » 

  "Owner: MULTIVERSE COMPUTING, S.L., a company with registered offices in Donostia-San Sebastián (Gipuzkoa), Paseo Miramón, nº 170, 3º-03, registered in the Companies Register of Gipuzkoa in volume 2.894, folio 202, page SS-41.659 and with tax identification number B-75218040 ..."

Re: Quasar 438B: Europe's Leading AI Model

#99

Post by torginus »

yorwba wrote: post #81
PorciiVorbesc wrote: post #42

Earlier quoted context omitted.

>For instance, I heard one of the Chinese models has some interesting "history" "facts" built into it EU models will also have such guard rails built in to "protect" the EU population from "misinformation" and uncomfortable facts and the EU elite don't want publicly known/discussed. Like I saw a post from German EU MEP where they asked the internal on-prem AI they have there "how can you vote out out Ursula v.d. Leye…

The question was not about voting. It was about getting rid of her. There are ways to get rid of elected officials that don't involve voting. My guess is that the model started brainstorming assassination plans before it got interrupted by the safety filter.

The question was 'how can we get rid of Ursula von der Leyen as the president of the commission'

Interpreting this as an assassination plot is quite the leap of logic.

Edit: Oh,it's the AfD

Re: Quasar 438B: Europe's Leading AI Model

#100

Post by torginus »

walrus01 wrote: post #17
Schlagbohrer wrote: post #14

I am still very surprised how absent India is from the LLM game.

Individual Indians are very much present in companies like OpenAI, Anthropic, etc. I have a theory that the top 5% talent of Indian nationals who are legitimately qualified and skilled, for very logical reasons would much rather take a six-figure USD equivalent salary in EUR or USD and enjoy the lifestyle benefits that come with it rather than bootstrap an Indian domestic AI lab.

It's not like people in other places couldn't solve this problem. Yes, the US pays more, but a lot of people would still prefer staying home, and it's not like they wouldn't be paid very well compared to the local salary.

And besides, a lot of Indians in India are working for FAANG, and don't earn US money - I'm not sure why it wouldn't make sense for some of these talented engineers to work for an indigenous tech company.

Post reply on HN