🔗 Linear Layer Visualizer
Every connection's thickness and color IS its weight —
green
for positive,
red
for negative.
📋 How To Use This Lab
Drag
in_features
/
out_features
— the whole bipartite graph rebuilds live, animated.
A
thick green
line means a strong positive weight; a
thick red
line means a strong negative one.
Flip
Bias
on — every output gets a "+b" badge and its value shifts accordingly.
Use
Next/Previous/Auto Run
to jump between a few interesting layer shapes.
Wira
Input
Output
◀ Previous Step
Next Step ▶
Auto Run ▶▶
↺ Reset
🤖 WIRA
Drag a slider — I'll rebuild every connection and recompute every output live.
🧑🏫 PROF. TORCH
nn.Linear is just a weight matrix: each output is a weighted sum of every input, plus an optional bias.
🎛️ Layer Shape
in_features:
3
out_features:
4
No Bias
Bias On