Selvauma/eas-mlip-training-data

Engineered Atomic Structures (EAS) — Training Data for Complex MLIPs The training dataset behind EAS-MLIP: a foundation machine-learned interatomic potential for structures engineered from distinct fundamental units — crystals, surfaces, molecules, electrolytes, monomers, polymers, and their interfaces — rather than one fixed material class. This repo is the data; the model itself (currently training, 4-seed MACE ensemble) will follow as a separate release once it converges.… See the full description on the dataset page: https://huggingface.co/datasets/Selvauma/eas-mlip-training-data.