What is a primary characteristic of non$-$parametric learning algorithms like kNN$?$

They do not assume a predetermined form for the model

They typically involve a simpler computation than parametric models

They assume a fixed form for the model

They are less susceptible to overfitting