In pipeline.config, there is a train_config section that appears to have some great data_augmentation_options that I normally do prior to
pipeline.config
train_config
data_augmentation_options