Abstract
Tuning machine learning models at scale, especially finding the righthyperparameter values, can be difficult and time-consuming. In addition to thecomputational effort required, this process also requires some ancillaryefforts including engineering tasks (e.g., job scheduling) as well as moremundane tasks (e.g., keeping track of the various parameters and associatedresults). We present Auptimizer, a general Hyperparameter Optimization (HPO)framework to help data scientists speed up model tuning and bookkeeping. WithAuptimizer, users can use all available computing resources in distributedsettings for model training. The user-friendly system design simplifiescreating, controlling, and tracking of a typical machine learning project. Thedesign also allows researchers to integrate new HPO algorithms. To demonstrateits flexibility, we show how Auptimizer integrates a few major HPO techniques(from random search to neural architecture search). The code is available athttps://github.com/LGE-ARC-AdvancedAI/auptimizer.