Abstract: Argonne National Laboratory is a scientific hub with initiatives ranging from cancer research to quantum chemistry to cosmology. Today, I will talk about my experience as a postdoctoral researcher and early-career computational scientist by exploring three major projects that I spend my time with: AuroraGPT, and the effort to train and develop trillion-parameter scale models with open methods and tailored to scientific use, GenSLM, the project to create language models for biological sequences such as peptides and genomes, and finally, I will talk about my work with the Hybrid-hardward Advanced Cosmology Code (HACC) team to generate domain-specific foundation models for cosmology, astrophysics and astronomy.
Bio: Azton Wells is an expert in the development and application of deep learning methods in scientific applications with a current focus on transformer-based models within high-performance computing environments. He received his PhD in Physics from the University of California, San Diego, and has since worked to apply his ML/AI expertise to domains as varied as language, genomics, and cosmological simulations. He currently holds a position as Assistant Computational Scientist in the Computational Sciences Division at Argonne National Laboratory focusing on building the AI initiative in computational sciences through leadership of the Frontier AI Task Force and the post-training team of the AuroraGPT project.