Fetching the paper…

Variance Reduction for Policy Gradient with Action-Dependent Factorized Baselines · Around