This is specific to manual experiments in building a deep learning model for finding good hyperparameters. What are the heuristics you use and any good resources available online?