Hey all, a team I’m part of just open-sourced our internal HPO tool called Auptimizer. Auptimizer does a couple of things. It provides a single interface to 6 different HPO algorithms including Spearmint and HyperOpt. It also makes it easy to scale your model training from CPUs and GPUs all the way to multiple instances on AWS. The repo is on Github. We have an article about it on Medium and you can find more implementation details in our 2019 IEEE Big Data paper. If you do HPO, check it out and let us know on Github how things look.