SGD Classifier and Regressor (page 4 of 9)

The step size (slop of the tangent line multiplied by the learning rate) matters when finding the minimum. The algorithm stops when the step size multiplied by learning rate is less than a specified value or a maximum number of iterations is reached.