Gradient Descent

Drag the starting point, tune the learning rate and momentum, and watch optimization either settle into a basin or overshoot. Inspired by Stanford CS229. Canvas 2D · machine learning.