Run ML workloads seamlessly on cloud TPUs and GPUs with a single Python decorator. No infrastructure management required. - View it on GitHub