Gradient Descent Question

Hi, so I am pretty new to ML and for my first project, my teacher wanted me to learn the basics of linear regression using stochastic gradient descent. My question is this: when you are updating each new theta, how do you calculate alpha, the learning rate?