Many-body-accurate Electrochemical Surface Diagrams Beyond Equilibrium

The dataset includes the training and test sets for the PBE, RPBE, and Δ-learning models, together with the corresponding trained models. The PBE and RPBE models were trained using the NequIP MLIP framework, while the Δ-learning model was trained using n2p2. In addition, the structural ensembles generated from sampling based on PBE, RPBE, and RPA energetics are provided.