Applying autoregressive neural networks to the temporal point process (TPP) framework has become the de facto standard for modeling continuous-time event data. Even though these models are expressive in modeling event sequences in a one-step-ahead fashion, they are inherently limited for long-term forecasting applications due to the accumulation of errors caused by sequential sampling.By deriving a probabilistic diffusion model for TPPs, we propose ADD-THIN, a new framework that naturally handle…