Gradient Descent Playground

4190.428 · Introduction to Machine Learning · Lecture 2
current θ steps taken planned path ✕ minimum θ*
Click the surface to set a start point · Space step · P play · R reset · N new landscape

Controls

0.15
30
0 / 30

Loss over steps

State

θ₁
–
θ₂
–
J(θ)
–
‖∇J‖
–
θ ← θ − α ∇J(θ)