A project is trying to cut the cost of making machine learning applications for Nvidia hardware, by developing on an Apple Silicon Mac and exporting it to CUDA. Machine learning is costly to enter, in ...