Matbench Discovery

Matbench Discovery is a benchmark dataset that contains a variety of materials science tasks used to evaluate the performance of machine learning models in predicting material properties.

Papers using Matbench Discovery (7)

AlphaNet: Scaling Up Local-frame-based Atomistic Interatomic Potential2025 · 7 citesDPA4: Pushing the Accuracy-Cost Frontier of Interatomic Potentials with EMFA SO(2) Convolution2026A Cartesian-3j Framework for Machine Learning Interatomic Potentials2025CrystalREPA: Transferring Physical Priors from Universal MLIPs to Crystal Generative Models2026EquiformerV3: Scaling Efficient, Expressive, and General SE(3)-Equivariant Graph Attention Transformers2026MatRIS: Toward Reliable and Efficient Pretrained Machine Learning Interatomic Potentials2026Orb: A Fast, Scalable Neural Network Potential2024 · 47 cites