Building some tests so that various approaches to solving continual learning, across different softwares, can be effectively and fairly compared.