Deep learning methods are transforming research, enabling new techniques, and ultimately leading to new discoveries. As the demand for more capable AI models continues to grow, we are now entering an era of Trillion Parameter Models (TPM), or models with more than a trillion parameters -- such as Huawei's PanGu-Σ. We describe a vision for the ecosystem of TPM users and providers that caters to the specific needs of the scientific community. We then outline the significant technical challenges and open problems in system design for serving TPMs to enable scientific research and discovery. Specifically, we describe the requirements of a comprehensive software stack and interfaces to support the diverse and flexible requirements of researchers.
@article{arxiv.2402.03480,
title = {Trillion Parameter AI Serving Infrastructure for Scientific Discovery: A Survey and Vision},
author = {Nathaniel Hudson and J. Gregory Pauloski and Matt Baughman and Alok Kamatar and Mansi Sakarvadia and Logan Ward and Ryan Chard and André Bauer and Maksim Levental and Wenyi Wang and Will Engler and Owen Price Skelly and Ben Blaiszik and Rick Stevens and Kyle Chard and Ian Foster},
journal= {arXiv preprint arXiv:2402.03480},
year = {2024}
}
Comments
10 pages, 3 figures, accepted for publication in the proceedings of the 10th IEEE/ACM International Conference on Big Data Computing, Applications and Technologies (BDCAT2023)