Python python toolset for Structure-Informed Property and Feature Engineering with Neural Networks. It offers unique advantages through (1) effortless extensibility, (2) optimizations for ordered, dilute, and random atomic configurations, and (3) automated model tuning.
See https://github.com/PhasesResearchLab/pySIPFENN/issues/18