Argonne National Laboratory and the Japanese national scientific research institute Riken have agreed to cooperate on a project focused on training large language models on scientific data.
Under the AuroraGPT project, ANL and Riken will collaborate with researchers from the Trillion Parameter Consortium to identify solutions to the challenges of large-scale AI system development.
The research organizations sealed the partnership with the signing of a memorandum of understanding at a virtual event
ANL Director Paul Kearns and Riken President Makoto Gonokami sealed the partnership in a virtual signing event on April 5.
As part of the cooperation, ANL and Riken will build and assess new computing system software and AI science and engineering applications, including generative AI models, computing system operations and data management.
According to Gonokami, researchers today have the responsibility to ensure that generative AI will positively contribute to humanity.
For Kearns, the collaboration opens up opportunities for significant advances for both sides.
“We have formed a partnership that can make pivotal discoveries and transformative advancements. We’re addressing some of the most complex challenges facing society, and I’m confident that we can achieve an even bigger impact by working together,” the director said.
The agreement includes sharing data, conducting lectures, joint seminars and symposia, and exchanging researchers, staff and students.