Bringing Atomistic Deep Learning to Prime Time

A common thread in headline-grabbing AI achievements is the need for massive compute resources, which are often only available to large corporations. Recent reviews of large language models have argued that non-profit research organizations should also be investigating models at scale [3]. Even doing a thorough hyperparameter optimization for a moderately-sized dataset can be a burden. As the business value of atomistic DL grows, large tech companies will become increasingly active in the field.