Source code and settings sufficient to reproduce benchmarks reported in the paper, "Smooth, exact rotational symmetrization for deep learning on point clouds." The lightweight file "source_code_settings.zip" contains 1) the source code of PET, 2) proof-of-principle implementation of ECSE, 3) all the settings for training procedures as a collection of YAML files with hyperparameters, 4) similarly organized parameters of ECSE, and 5) a definition file of the Singularity container (.def) we used for most of the experiments. In addition to the above, "Source_code_settings_checkpoints.zip" contains 1) all the checkpoints, including all intermediate ones for same calculations, 2) the Singularity container (.sif) itself, and 3) the datasets we used.