Direct Solution vs ML Libraries — How Do They Compute w? In this short, we compare the Direct Solution for linear regression with how TensorFlow and Scikit-learn compute the weight vector w. TensorFlow’s Dense layer learns the bias internally, Scikit-learn keeps the code simple, and both frameworks use optimization — not the Normal Equation — to find w. If you want to see the full breakdown of Direct Solution vs Gradient Descent, tap below for the full lecture. ==================================