Training and Test Datasets for a MACE Interatomic Potential for SiO2

This dataset contains the training and test configurations used to develop a MACE machine-learning interatomic potential for SiO2. It includes several crystalline SiO2 polymorphs and combines equation-of-state, phonon-displacement, and randomly distorted configurations calculated using density functional theory. The validation subset is generated internally from the training file during model training.