Forward & Backpropagation Visualizer
Trains the 2-input, 2-hidden-unit, 1-output network on a single example (x1, x2) → y_true, one gradient descent step at a time. Click
Step
to walk through each phase by hand, or
Run
to watch it repeat and converge automatically.
x1
1.20
x2
-0.60
y_true
0.85
Learning rate α
0.15
Speed
3×
▸ Step
▶ Run
⏸ Pause
↻ Reset
Ready — click Step or Run to begin.
Input
Hidden
Output
Weight / bias
Gradient (backward)
Step
0
Loss L
—
y_pred
—
y_true
0.85
Ready — click Step or Run to begin.