When the target population differs from the experiment sample, data practitioners must apply weighting estimators for the population average treatment effect (PATE) to generalize the causal impact of a treatment. However, most methods for estimating the PATE are insufficient when the experiment sample and target population are very different. Here, Huang et al. propose using outcome data from the target population to reduce noise in generalizing experimental results. They present a post residualized weighting technique wherein they train an ML model on observational data to predict outcomes and then use this model to residualize the experimental outcome. The resulting residuals replace the experimental outcome data when implementing conventional weighting methods for generalization. The authors conclude by using simulations to evaluate the performance of the proposed post-residualized weighting estimators.