Fig. 12From: Scalable training of trustworthy and energy-efficient predictive graph foundation models for atomistic materials modeling: a case study with HydraGNNWeak scaling. AMD is better for large modelsBack to article page
