A comparison of deep learning frameworks has been made, with one framework reportedly having an 85% research share and a 10% training speed gap. Other comparisons of programming tools and libraries have also been conducted, including those for data analysis and user interface development. The specifics of these comparisons and their findings are varied and not entirely clear.