MAD

The MAD dataset is a highly curated collection of atomistic configurations, including molecules, clusters, bulk crystals, surfaces, and low-dimensional structures, designed to evaluate and improve the training of broadly applicable machine learning models across the periodic table using consistent DFT calculations.

Papers using MAD (3)

Massive Atomic Diversity: a compact universal dataset for atomistic machine learning2025 · 11 citesHigh-quality, high-information datasets for universal atomistic machine learning2026 · 1 citesFlexible Cutoff Learning: Optimizing Machine Learning Potentials After Training2026