Back to Tools

Optimizer Simulator

Visualize how different optimization algorithms navigate loss landscapes.

Landscape

Curvature Control

X Curvature (a)1.0
Y Curvature (b)10.0

High ratio creates a narrow valley (bad condition number).

f(x,y)=ax2+by2f(x,y) = ax^2 + by^2

Hyperparameters

Learning Rate0.01
Momentum β\beta0.9
Adam β1\beta_10.9
Adam β2\beta_20.999
Loading Optimizer Visualization...
Iterations: 0
Insight:
Notice how Adam and RMSProp adapt their step sizes to the curvature (moving fast in flat directions, slow in steep ones), often converging faster on the Quadratic Bowl with high curvature. However, Momentum builds up velocity and can oscillate or overshoot, but often escapes shallow local minima better than pure SGD.