Switzerland Develops Open-Source Artificial Intelligence Language Model

Switzerland has developed an artificial intelligence language model with a completely open-source code structure. The model, not yet publicly accessible, was trained with a dataset covering over 1,500 languages. What the model will offer is already a matter of great curiosity.

A European country has joined the AI wars we typically see between the US and China. This country, surprisingly, is Switzerland. An open-source AI language model has been developed through a collaboration between ETH Zurich University, EPFL University, and the Swiss National Supercomputing Centre (CSCS). This public-interest endeavor seems poised to usher in a new era in the AI sector.

The language model developed within the scope of the work in Switzerland was trained with a massive dataset encompassing over 1,500 languages. The model, which is compliant with the European Union AI Act, will be made accessible to everyone via an Apache license. According to the statement, Switzerland’s AI language model will be released in two versions, featuring 8 billion and 70 billion parameters.


NVIDIA Chips Used!

Switzerland’s developed AI model was created on Alps supercomputers, produced by CSCS and operating with carbon neutrality. The scientists, who used over ten thousand NVIDIA Grace Hopper chips for the model’s development, believe this model will be rapidly adopted in science, government affairs, education, and other industry sectors.

Generally, AI technologies are developed behind closed doors and for profit. No one truly knows what companies like OpenAI, Microsoft, and Google are doing in the background. Switzerland’s adopted method, however, is quite remarkable as it demonstrates that transparency can also be a core principle in the AI sector, and that everything isn’t about money. It remains to be seen how the performance of the model, planned for release in the coming period, will be.

You Might Also Like;

Exit mobile version