When creating a post, please add:
For week 2 lab 1 in visualizing the gradient descent for one variable, how is the convergence criteria being verified? I see the function taking all the inputs (function ‘gradient_descent_one_variable’) and calculating the points over iterations. But how is it deciding on whether the point has converged or not? I am asking from both, the code perspective and theoretical perspective.
Convergence means that the cost has reached a stable minimum. In practice that means it’s close enough that the gradients are near enough to zero as makes no difference.
1 Like
Understood. Thank you.