Deep Learning for Science School

The Deep Learning for Science Summer School has been hosted by Computing Sciences at Berkeley Lab in 2019 and 2020. The school focuses on teaching attendees about applying state-of-the-art deep learning to scientific applications and best practices for running these models on high-performance computing systems.

The 2025 Deep Learning for Science (DL4SCI) Summer School restarts this event and is a five-day immersive program bringing together researchers and engineers to explore the latest advances in deep learning and generative AI (GenAI). 

The school will feature in-depth lectures, research talks, and hands-on tutorials on state-of-the-art techniques, including transformers, large language models, diffusion models, flow matching, and generative adversarial networks, as well as practical sessions on best practices for running deep learning at scale on high-performance computing systems. Theory will be covered, but the program places an emphasis on practical application. Local and international AI experts will lead interactive labs and offer state-of-the-art code and content. Attendees will gain a strong understanding of what GenAI is, the types of problems it excels at, and how to choose, build, train, and deploy GenAI models for scientific applications.

For more information and to apply, please visit the school webpage.

For older schools, see 2019 summer school material and 2020 summer school material.