← AI & ML
Gradient Descent Visualiser
UK
Optimiser
SGD
Step
0
Loss
—
‖∇‖
—
Δ pos
—
Global min
—
Converged?
—
Optimiser
SGD
SGD + Momentum
RMSprop
Adam
All four (compare)
Learning Rate
0.05
Momentum / β₁
0.90
Gradient Noise
0.10
Loss Function
Rosenbrock (banana)
Saddle point
Beale's function
Himmelblau (4 minima)
↺ Reset
⏸ Pause
Optimisers
SGD
Momentum
RMSprop
Adam
Drag to rotate · Scroll to zoom