4D Parallelism Unlocks Exascale Bayesian Neural Networks for High-Fidelity Atmospheric Modeling
An orthogonal 4D-parallelization scheme is devised that introduces a unique domain-tensor-parallelism strategy and a novel uncertainty parallel method, enabling the potential of high-fidelity uncertainty quantification in atmospheric AI models, heralding a new era for AI-based models in climate and Earth system science...