In this paper, we propose a novel predictive safety filter that is robust to bounded perturbations and is combined with a learning-based control called differentiable predictive control (DPC). The proposed method provides rigorous guarantees of safety in the presence of bounded perturbations and implements DPC so long as the DPC control satisfies the system constraints. The approach also incorporates two forms of event-triggering to reduce online computation. The approach is comprised of a robust predictive safety filter that extends upon existing work to reject disturbances for discrete-time, time-varying nonlinear systems with time-varying constraints. The safety filter is based on novel concepts of robust, discrete-time barrier functions and can be used to filter any control law. Here we use the safety filter in conjunction with DPC as a promising policy optimization method. The approach is demonstrated on a single-integrator, two-tank system, and building example.