Understanding Apertus, the Swiss LLM research platform

This workshop introduces the Apertus LLM, an open-source large language model developed through the Swiss National AI Initiative (SNAI). We begin with a 20-minute overview highlighting its Swiss origins, collaboration within SNAI, and its transparency and open-source nature. The model's world-leading multilingual capabilities, technological innovation, and legal compliance will be discussed, with a focus on data privacy and sovereignty.

The 45-minute deep dive session provides hands-on experience with Apertus, covering topics such as getting started, benchmarking, prompting, and exploring its language capabilities. A significant portion will focus on sustainability, discussing training in climate-neutral data centers, energy consumption monitoring, and strategies to minimize energy usage during inference. Practical examples and recent research will illustrate the model's impact on sustainable practices. We will split into beginner and advanced groups depending on participants' skill levels.

The workshop concludes with a joint Q&A session, addressing questions about carbon costs, cloud usage, language benchmarks, system integration, and compliance with Swiss privacy laws. Attendees will be encouraged to explore Apertus further via the Public AI Platform, academic or commercial APIs. We will collect feedback for ongoing development of the LLM.

Join us to learn how Apertus can support your AI sustainability roadmap and contribute to the future of Swiss AI!