Across all functionals, we observe a clear trend of decreasing MAPE as the basis set quality increases: for both uncorrected DFT ( Figure a) and ML-corrected LR predictions ( Figure b), MINIX exhibits the largest errors and def2-QZVP the smallest.
← all excerpts
Transferable and Transparent Energy Decomposition-Based Machine Learning Models for Computing Accurate Reaction Energetics.
1
—
—