A novel class of MTPP models inspired by linear Hawkes processes and deep state-space models, that brings linear complexity and sublinear scaling while being highly expressive.

We propose a unified transformer framework to model mixed-type event sequences