Switzerland’s top technical universities, ETH Zurich and EPFL, are set to develop a public, open-source large language model (LLM) by 2025, emphasizing ethics and scientific integrity. Under the Swiss AI Initiative, launched in December 2023, over 10 Swiss institutions will collaborate, coordinated by the Swiss National AI Institute and involving 70 AI researchers from ETHZ and EPFL. The model will be trained on the Alps supercomputer using over 10,000 NVIDIA Grace Hopper chips. Unlike many proprietary LLMs like ChatGPT and Claude, this Swiss model prioritizes transparency, crucial for gaining trust in Europe amidst concerns over digital sovereignty. It will support over 1,000 languages, with a training dataset containing content in over 1,500 languages, ensuring global applicability. This initiative aims to foster innovation and accountability across science, government, and education sectors, paving the way for high-trust AI applications.
For more updates on Swiss technology, follow us on Facebook and Twitter.
Source link
