Our schedule of training events for all AI communities in the UK, including all ExCALIBUR projects, academic institutions, and public sector research establishments, and Community Workshops open to industry, academia, and national laboratories, is currently being finalised.

Information will be shared here, via the ExCALIBUR network and our social media channel. If you are interested to hear our updates, follow us on LinkedIn!

Benchmarking LLMs on large-scale systems

7 February 2024, 4-5pm GMT

 Join Ana Gainaru (ORNL) as she shares her experience with optimizing LLM performance at scale and a roof model for predicting the performance of the LLM’s performance when deployed on ORNL’s Frontier supercomputer, and will focus on large complex workflows that couple LLMs with traditional HPC simulations when the performance bottleneck migrates from execution to data management. Explore novel strategies for mitigating this new challenge and ensuring seamless information flow within such hybrid workflows. Register now to join Ana on this journey!

Your content goes here. Edit or remove this text inline or in the module Content settings. You can also style every aspect of this content in the module Design settings and even apply custom CSS to this text in the module Advanced settings.

Deep Learning - Neural Networks: The Challenge of Extrapolation for Deep Neural Operators

9 May 2024, 4-5pm BST

Join Lu Lu  (Yale University) as he shares how deep learning has achieved remarkable success in diverse applications; moreover, the emerging scientific applications. Reviewing physics-informed neural networks (PINNs) and available extensions for solving forward and inverse problems of partial differential equations (PDEs), Lu continues to introduce a less known but powerful result that a NN can accurately approximate any nonlinear operator. This universal approximation theorem of operators is suggestive of the potential of NNs in learning operators of complex systems. Lu will present the deep operator network (DeepONet) to learn various operators that represent deterministic and stochastic differential equations, and demonstrate the effectiveness of DeepONet and its extensions to diverse multiphysics and multiscale problems, such as nanoscale heat transport, bubble growth dynamics, high-speed boundary layers, electroconvection, hypersonics, and geological carbon sequestration. Deep learning models are usually limited to interpolation scenarios, and Lu will quantify the extrapolation complexity and develop a complete workflow to address the challenge of extrapolation for deep neural operators.

Join us for this technical yet practical exchange.

Past Events Archive

We are extremely thankful to all our speakers for contributing their time and expertise to our Knowledge Exchange events.

Never miss out on one of our exciting talks! As we finalise each talks video, you will be able to watch it again and access any slides or distributed materials here.