These two methods accept an function that operates on a Grid object to do additional customization of the plot. The methods are very similar, but apply returns the Grid it is attached to while pipe returns the value from the passed function. The objective is to make it easier to do arbitrary customization without catching the return value and switching to an imperative programming style.

Example:

def slopeline(grid, slope, **kwargs):
    grid.ax.axline((0, 0), slope=slope, **kwargs)

(
    sns.relplot(tips, x="total_bill", y="tip")
    .apply(slopeline, slope=.2, color=".6")
    .set_axis_labels("Total bill", "Tip")
)

Additionally, fixed two methods (tight_layout and refline) that were not returning self.