R/optimizer.R
sgd.Rd
Simple stochastic gradient descent without momentum or preconditioning.
sgd(stepsize = 0.001)
stepsize for SGD
a list of control variables for optimization (used in control_opt function)
control_opt
