We share the crucial DFT results for the structures in the reference dataset used to train the neural network potential. These are available in 'CuW_reference_dataset_Quantun_ESPRESSO.zip'. Each subfolder contains an input file and an output file (input file (QE.in) and output file (QE.out)) for one structure. The simulations were carried out with the Quantum ESPRESSO package. Subfolders are named with the convention: ___ Arguments for each property are as follows: - EOS - _ or _ - elastic - _ - n and p represent negative and positive strains respectively - decohesion - _ - interstitial - - tetrahedral or octahedral - stacking-fault - __ - stacking-fault-surface - __ - interstitial-transition - - vacancy - no keywords - surface - - atom-substitution - - interface - - The relaxed interface structure - atomic-swap-interface - __ - atoms swapped are described as the neighboring atom relative to the interface (i.e. 1NN, 2NN, or 3NN), relaxation status can be unrelaxed, fully relaxed, and z-relaxed (atoms are only allowed to displace normal to the interface plane) - interface-vacancy - __ - slip-energy-surface - __ - random-displacements - _dx__ - relaxation-intermediate - _ - NNP-with-errors - - Structures with substantial errors for energy/forces from previous training runs *The space group is given for phrases that don't have a common notation **Interface notations are defined in the manuscript ***Not all the intermediate relaxation structures are included. The number is based on the number of structures included per each configuration The input.data file was created from the structures in the reference dataset for units angstrom (distance), eV/angstrom (forces), and eV (energies). The energies of isolated CU and W atoms were deducted from the total energy of each structure.