On the Utility of Equal Batch Sizes for Inference in Stochastic Gradient Descent

Re-assign community