Not sure this is too interesting, but I however think that the case when $X \rightarrow \text{Spec}(\mathbb{Z})$ is proper and smooth(!) should follow from a result proven independently by Fontaine and Abrashkin that implies that the Hodge-numbers are zero for $i+q \leq 3$ and $i \neq j$.