The data in this repository support the findings presented in the article "Atomistic modeling of bulk and grain boundary diffusion in solid electrolyte Li6PS5Cl using machine-learning interatomic potentials" by Ou et al. The repository contains the training sets, the fitted machine-learning interatomic potentials (MTPs), and the relaxed bulk and grain boundary structures. An automated script to perform the proposed quality-level-based active learning scheme is also provided.