Supplement to: User-Dependent Neural Sequence Models for Continuous-Time Event Data A Model Details A.1 Encoder Details

The embedding for the mark is a standard, learned embedding. This transformation is the same as the positional embeddings from V aswani et al. Figure A: A detailed operational diagram of the encoding process. An operational diagram of the encoder, as described previously, can be found in Figure A. A graphical MTPPs, as events for each mark can be seen as coming from their own point process. We utilize this property in a thinning procedure to sample from arbitrary point processes. Table A contains model hyperparameters used for all of the experiments.