Median relative error of the interpolation expressed as a percentage (left) and the fraction of relative errors above 10% (right) from all iterations of the 50 leave 5% of the sample out tests for six interpolated quantities. The points on each plot, moving from left to right, represent the different iterations; we exclude all simulations that stopped due to initial Roche-lobe overflow (indicating a difference between COMPAS and MESA models). The right plot includes NaNs (obtained from non-fittable points, e.g., points at the boundary of the parameter space) when counting relative errors larger than 10%, while the left plot excludes them.