4 ms·Exascale Training of Generative Model with Historical Priors for Data Reduction1 points by rbanffy 6mo ago