Language models are taking off
Language models are taking off
The NorwAI language model activities received national attention in 2023. The interest continues into 2024 with VIP political visits.
August 2023
Our public introduction to present Norwegian Large Language Models took place in ‘Arendalsuka’ on 17th of August 2023. Together with Sven Størmer Thaulow from Schibsted and Yngvar Ugland from DNB NewTechLab the team from NTNU presented our largest language model, the NorGPT-23B generative model. Present were also two representatives from the Ministry of Local Government and Regional Development, minister Sigbjørn Gjelsvik and state secretary Gunn Karin Gjul. The discussion afterwards was very encouraging, and there was a clear understanding that Norwegian-based language models are needed.
September/October 2023
Tore O. Sandvik from the Labour Party, now state secretary of the Trade and Industry Ministry, visited us on 14th of September. Mr Sandvik also gave a presentation at our our national AI conference, the NorwAI Innovate, on trustworthiness of technology as October shifted to November.
The first day of the conference paid considerable attention to language models. Professor Fredrik Heintz of Linköping, coordinator of the EU language project “Trust LLM” held a talk on ‘Towards Factual and Trustworthy Lange-Language Models’ while Per Kummervold at the National Library of Norway talked about the Library’s work. The industrial perspectives were presented by Daniel Pleus, Data Scientist Lead at Schibsted, with a talk about the media frontrunner’s work on language models, while Jon Espen Ingvaldsen of NorwAI updated the audience on NorLLM.
November 2023
The new minister of Digitalization and Public Governance, Karianne Tung, came to Gløshaugen on 14th of November to further discuss the prospects of training large-scale Norwegian language models. These discussions have of course not been limited to government parties, and Venstre with Guri Melby and Grunde Almeland paid us a visit on 6 th of December.
December 2023
At the end of the year both NTNU and UiO were invited by the National Library to investigate the value of copyrighted material in large-scale language models. This was an initiative from the Ministry of Culture and Equality, and the hope is that this work will lead to larger and higher-quality training datasets for even better language models in the future.
February 2024
The political interest for national language models continued into the new year. NorwAI made a pre-presentation on the next generation of models in Trondheim in mid February. On February 26th the minister of Minister of Research and Higher Education Oddmund Hoel came to visit NorwAI displaying deep insights into central questions and dilemmas of the speedy developments of large language models.
The interest continues. On May 15th NorwAI will present four new language models, and plan to update our audience further at the upcoming ‘Arendalsuka’ in August.