ORNL Joins Other National Labs in Global Consortium Solving Large AI Problems

Oak Ridge National Laboratory has joined the Trillion Parameter Consortium, an organization of scientists that addresses challenges with large-scale artificial intelligence systems and advances AI capabilities for scientific discoveries.

ORNL will help the consortium build a community interested in generative AI model development, launch and coordinate projects that would benefit the broader AI and scientific community and create a global network of resources and expertise for next-generation AI. The laboratory will also join TPC’s efforts in developing scalable AI model architectures, curating and organizing AI model training modules, optimizing AI libraries for current and future exascale computers and assessing the progress of scientific activities.

Other Department of Energy national labs are part of the consortium, in addition to laboratories from Europe and Asia, ORNL said Monday.

The laboratory is home to Frontier, an exascale computer considered the fastest of its kind when it was installed in 2022. According to Doug Kothe, associate director of ORNL’s Computing and Computational Science Directorate, Frontier can enable unprecedented machine learning, automation and data processing, which can improve the United States’ understanding of nuclear energy, climate and other applications.

Frontier can perform at 1.1 exaflops per second and around 63 billion calculations per watt.