If the King of Sweden wants help drafting his annual Christmas speech this year, he could ask the same AI model that's available to his 10 million subjects.
As a test, researchers prompted the model, called GPT-SW3, to draft one of the royal messages, and it did a pretty good job, according to Magnus Sahlgren, who heads research in natural language understanding at AI Sweden, a consortium kickstarting the country's journey into the machine learning era.
Later, our minister of digitalization visited us and asked the model to generate arguments for political positions and it came up with some really clever ones - and he intuitively understood how to prompt the model to generate good text, Sahlgren said.
Early successes inspired work on an even larger and more powerful version of the language model they hope will serve any citizen, company or government agency in Scandinavia.
A Multilingual Model The current version packs 3.6 billion parameters and is smart enough to do a few cool things in Swedish. Sahlgren's team aims to train a state-of-the-art model with a whopping 175 billion parameters that can handle all sorts of language tasks in the Nordic languages of Swedish, Danish, Norwegian and, it hopes, Icelandic, too.
For example, a startup can use it to automatically generate product descriptions for an e-commerce website given only the products' names. Government agencies can use it to quickly classify and route questions from citizens.
Companies can ask it to rapidly summarize reports so they can react fast. Hospitals can run distilled versions of the model privately on their own systems to improve patient care.
It's a foundational model we will provide as a service for whatever tasks people want to solve, said Sahlgren, who's been working at the intersection of language and machine learning since he earned his Ph.D. in computational linguistics in 2006.
Permission to Speak Freely It's a capability increasingly seen as a strategic asset, a keystone of digital sovereignty in a world that speaks thousands of languages across nearly 200 countries.
Most language services today focus on Chinese or English, the world's two most-spoken tongues. They're typically created in China or the U.S., and they aren't free.
It's important for us to have models built in Sweden for Sweden, Sahlgren said.
Small Team, Super System We're a small country and a core team of about six people, yet we can build a state-of-the-art resource like this for people to use, he added.
That's because Sweden has a powerful engine in BerzeLiUs, a 300-petaflops AI supercomputer at Link ping University. It trained the initial GPT-SW3 model using just 16 of the 60 nodes in the NVIDIA DGX SuperPOD.
The next model may exercise all the system's nodes. Such super-sized jobs require super software like the NVIDIA NeMo Megatron framework.
It lets us scale our training up to the full supercomputer, and we've been lucky enough to have access to experts in the NeMo development team - without NVIDIA it would have been so much more complicated to come this far, he said.
A Workflow for Any Language NVIDIA's engineers created a recipe based on NeMo and an emerging process called p-tuning that optimizes massive models fast, and it's geared to work with any language.
In one early test, a model nearly doubled its accuracy after NVIDIA engineers applied the techniques.
Magnus Sahlgren What's more, it requires one-tenth the data, slashing the need for tens of thousands of hand-labeled records. That opens the door for users to fine-tune a model with the relatively small, industry-specific datasets they have at hand.
We hope to inspire a lot of entrepreneurship in industry, startups and the public using our technology to develop their own apps and services, said Sahlgren.
Writing the Next Chapter Meanwhile, NVIDIA's developers are already working on ways to make the enabling software better.
One test shows great promise for training new capabilities using widely available English datasets into models designed for any language. In another effort, they're using the p-tuning techniques in inference jobs so models can learn on the fly.
Zenodia Charpy, a senior solutions architect at NVIDIA based in Gothenburg, shares the enthusiasm of the AI Sweden team she supports. We've only just begun trying new and better methods to tackle these large language challenges - there's much more to come, she said.
The GPT-SW3 model will be made available by the end of year via an early access program. To apply, contact francisca.hoyer@ai.se.
Most recent headlines
04/09/2025
Monumental Sports & Entertainment (MSE), in collaboration with Dalet, has been a...
23/04/2025
In 2024, Thomson's Total Turnout project played a pivotal role in elevating ethical election reporting and strengthening journalist safety in Pakistan.
Wit...
23/04/2025
We heard that you could use a little pick-me-up, so get ready to sh-sh-shake it ...
23/04/2025
WBD's Max viewership climbs 6%, boosted by The White Lotus' and The Pi...
23/04/2025
March shows a further downward trend of time spent watching television; following the February period when winter holidays contributed to a shorter time spent. ...
23/04/2025
During March, audiences in Mexico increased their streaming usage by 2.1 points compared to the previous month, accounting for 24.4% of TV viewing.
Disclaimer:...
23/04/2025
AANHPI audiences over index the total U.S. for share of time spent with Netflix ...
23/04/2025
SAN JOSE, Calif. Roku has announced new TVs, new streaming devices and significant upgrades to its user interface and software platforms that are designed to st...
23/04/2025
COW Job Listing: Opportunity for a Passionate Feature Film Editor - London-Based...
23/04/2025
RM Equity Partners Acquires MAGIX Software, Appoints Robert Rutkowski as CEO to ...
23/04/2025
All Men Are Wicked Western Shot with Blackmagic Design
Brie Clayton April 23, 2025
0 Comments
Blackmagic Pocket Cinema Camera 4Ks were put to the test...
23/04/2025
NEW YORK Warner Bros. Discovery captured the largest monthly viewership increase among media distributors in March, according to Nielsen's latest Media Dist...
23/04/2025
MIAMI, Fla. Hemisphere Media Group has inked a deal with Samsung to launch WAPA+ as a FAST channel on Samsung TV Plus, a free TV streaming service that comes pr...
23/04/2025
HEBDEN BRIDGE, U.K. Calrec Audio has promoted Sid Stanley to managing director. Stanley, who joined the company in July 2018 as general manager, has been instru...
23/04/2025
WASHINGTON Federal Communications Commission Commissioner Nathan A. Simington has announced a series of staff appointments made in March and April 2025, includi...
23/04/2025
WASHINGTON The Federal Communications Commission has set its agenda for the Monday, April 28, 2025 Open Meeting, which is scheduled to start at 10:30 a.m. in th...
23/04/2025
COW Job Listing: Full-Time Video Editor, Remote
Brie Clayton April 22, 2025
0 Comments
Full-Time Video Editor
April 23, 2025COW Job Listing: Opportu...
23/04/2025
Berklee Abu Dhabi's Jazz Night Kicks off Global Celebration of Music and Cul...
23/04/2025
The E.W. Scripps Co. said it was shutting down its Scripps News over-the-air channel effective November 15 and eliminating at least 200 jobs....
23/04/2025
Audio Helps Chess.com Checkmate Cheating Matches are produced using REMI workflows By Dan Daley, Audio Editor
Wednesday, April 23, 2025 - 7:00 am
Print Th...
23/04/2025
Tech Focus: Intercoms, Part 1 - Key to Onsite, REMI, Hybrid Operations Intercoms keep increasingly disparate production locations together By Dan Daley, Audio ...
23/04/2025
Speaking my language: Bringing AI to the fore with real time audio translation f...
23/04/2025
Back to All News
Netflix Drops Trailer for Crime Drama Secrets We Keep, Set in ...
23/04/2025
Back to All News
World Book Day: Netflix Announces New TV Adaptations and Impac...
23/04/2025
Comscore to Announce First Quarter 2025 Financial ResultsRESTON, VA, April 23, 2025 Comscore, Inc. (Nasdaq: SCOR), a trusted partner for planning, transacting...
23/04/2025
In an email to RT staff today, RT Director-General, Kevin Backhurst confirmed that RT 's Voluntary Exit Programme (VEP) 2025 is open from today, Wednesday...
23/04/2025
RT will air live coverage on television, radio and online of the funeral of Pope Francis which takes place this Saturday, 26 April 2025.
Beginning at 8.30am o...
23/04/2025
As AI models evolve and adoption grows, enterprises must perform a delicate balancing act to achieve maximum value.
That's because inference - the process ...
23/04/2025
Financial services has long been at the forefront of adopting technological innovations. Today, generative AI and agentic systems are redefining the industry, f...
23/04/2025
Facebook
Twitter
LinkedIn
Enables Michelin to focus on development of its ...
23/04/2025
AI is rapidly reshaping what's possible on a PC - whether for real-time image generation or voice-controlled workflows. As AI capabilities grow, so does the...
23/04/2025
An AI agent is only as accurate, relevant and timely as the data that powers it....
23/04/2025
April 23 2025, 02:00 (PDT) Dolby Ushers in the Future of In-car Entertainment at Auto Shanghai 2025
Dolby and industry partners demonstrate robust momentum ...
22/04/2025
Chloe Sevigny at the Magic Farm premiere (photo by Michael Hurcomb/Shutterstock for Sundance Film Festival)...
22/04/2025
Music has the power to connect us to one another and to the world around us. Spotify is once again teaming up with the Museum for the United Nations - UN Live&#...
22/04/2025
Spotify has an unwavering commitment to supporting emerging artists across all g...
22/04/2025
This year marks the 55th anniversary of Earth Day, celebrated annually on April 22. On Spotify, listeners can explore a world of stories about our planet in our...
22/04/2025
McAvaney takes the lead for SBS' Coverage of the World Athletics Championshi...
22/04/2025
Virtual idols are poised to revolutionize entertainment, with the market projected to top $4 billion by 2029. Crafted with motion capture and game engine techno...
22/04/2025
As the audiovisual (AV) industry rapidly evolves, educational institutions are preparing the next generation of AV professionals for the future. Singapore Polyt...
22/04/2025
DENVER A new survey highlights just how important streaming has become in the entertainment landscape. Data compiled by Adtaxi shows streaming isnt just an alte...
22/04/2025
Samsung TV Plus has announced that it is now offering nearly 700 streaming channels in the U.S. more than any of the other major FAST platforms, the streaming s...
22/04/2025
PORTSMOUTH, N.H. New findings from Hub Research's latest survey on sports and television coverage reveal Amazon may have cracked the code of how to use tech...
22/04/2025
DETROIT Rockbridge Growth Equity has sold a controlling stake in GSTV to MidOcean Partners. Rockbridge will retain a minority stake in the video network which i...
22/04/2025
ATLANTA Gray Media has promoted Dana Neves to senior managing vice president, overseeing a number of Gray's television markets in the Northeast and Mid-Atla...
22/04/2025
The Infrastructure of Creative Flow: DigitalGlue's creative.space Wins NAB 2...
22/04/2025
iZotope unveils Equinox: the ultimate reverb plugin for Post and Music productio...
22/04/2025
Every way to export transparent videos from After Effects
Graham Quince April 21, 2025
0 Comments
In this video, I break down each way you can export ...
22/04/2025
April 22nd, 2025 Press Materials Available Here
Tribeca Festival 2025 Announces Short Film Lineup
Featuring How I Learned to Die Executive Produced by Osca...
If the King of Sweden wants help drafting his annual Christmas speech this year, he could ask the same AI model that's available to his 10 million subjects.
