Hypergradient descent and Universal Probabilistic Programming

Online Learning Rate Adaptation with Hypergradient Descent: We introduce a general method for improving the convergence rate of gradient-based opti...