The tune package provides tools for hyperparameter tuning within the tidymodels ecosystem. It works with recipes for preprocessing, parsnip for model specification, and dials for defining parameter grids.
The package supports both grid search and iterative Bayesian optimization approaches to find optimal model parameters. It integrates with tidymodels workflows to evaluate different parameter combinations across resampling strategies like cross-validation. The package handles parallel processing for efficiency and provides acquisition functions for scoring parameter combinations during optimization.
