Gaussian Approximation and Concentration of Constant Learning-Rate Stochastic Gradient Descent | Read Paper on Bytez