Improving the description of multiple interactions in Herwig

The modeling of multiple parton interactions in Monte Carlo event generators is a crucial part not only for the dressing of signal processes but also to describe data with a minimum bias on the event selection. Much work has and will be put into the theoretical framework and the numerical implementation of these models. In this contribution, we document various improvements of the multiple parton interaction model of Herwig 7 (Bellm et al. in Eur Phys J C76(4):196, 2016), that lead to an improved description of minimum bias and underlying event data.