Efficient Machine Learning barrier height and reaction enthalpy corrections for low- and mid-level quantum mechanical calculations

This Zenodo repository contains the optimized molecular structures of stationary points (reactants, products, and transition states) used in the development of Δ-ML correction models. The archive includes four chemical databases — RDB7, GMTKN55, RGD1, and GLN — each organized by the baseline quantum mechanical method used for geometry optimization: PM6, HF/3-21G, and r²SCAN-3c. For each structure, the corresponding baseline method energy and the Δ-ML model predicted energy are provided. For further details on the models and computational methodology, please visit: https://github.com/ComputationalChem-USC/AMK-ML-corrections-BH