Text this: Improving drug discovery with a hybrid deep generative model using reinforcement learning trained on a Bayesian docking approximation.