Fetching the paper…

The Complexity of Making the Gradient Small in Stochastic Convex Optimization · Around