# Standard ![hpopt_overall](./image/hpopt_overall.png) - A user specifies the hyper-parameter search space. - hpopt create a new HpOpt object with the user-provided configurations. - hpopt tries to find the hyper-parameters that maximizes or minimizes the metric score by training a model with various combinations of hyper-parameters. - The best hyper-parameters is selected and is used by the following full fine-tuning phase.