Gradient Descent Explorer

Loss Surface

Learning Rate η

0.1500

Animation Speed

Live Stats

Iteration0

Loss L(w)—

Gradient |∇L|—

Position (w₁,w₂)—

Status○ Ready

Legend

Gradient ∇L — steepest ascent

Step −η∇L — where we move

Path history

★ Global minimum ● Current position

💡 Simple Bowl

A perfectly convex surface with one global minimum. Gradient descent always converges here — try different learning rates and watch the path change!