Calculating the loss function for every conceivable value of w over the entire data set would be an inefficient way of finding the convergence point. A better mechanism—very popular in machine learning—is called gradient descent.
Hyperparameters are the knobs that programmers tweak in machine learning algorithms. Most machine learning programmers spend a fair amount of time tuning the learning rate. If you pick a learning rate that is too small, learning will take too long
I will continue to work on Google's Machine Learning Crash Course