Reproducible Computational Research with Hydra

Configuration files, parameters, and output data are often scattered across multiple locations, making it difficult to reproduce results or determine how they were obtained. This micro-school introduces Hydra, a Python framework for managing experiment configurations and workflows. Participants will learn how to organise configurable applications, compose reusable configurations, and automatically record the settings used for each run. By adopting these practices, researchers can improve the reproducibility of their computational work, reduce errors caused by manual configuration management, and eliminate the frustration of losing experiment configurations.