Driving in dense traffic with human and autonomous drivers is a challenging task that requires high level planning and reasoning along with the ability to react quickly to changes in a dynamic environment. In this study, we propose a hierarchical learning approach that uses learned motion primitives as actions. Motion primitives are obtained using unsupervised skill discovery without a predetermined reward function, allowing them to be reused in different scenarios. This can reduce the total training time for applications that need to obtain multiple models with varying behavior. Simulation results demonstrate that the proposed approach yields driver models that achieve higher performance with less training compared to baseline reinforcement learning methods.