Chapters 21–25

CNN Architecture Lab

Configure CNN architectures, train on synthetic 8×8 patterns, and watch filters specialize in real time.

How to use: Choose a preset architecture or adjust filter counts. Pick a dataset (3 or 4 pattern classes), set the learning rate and number of epochs, then click Train. Watch the loss curve converge and inspect learned filters. Use the Inference tab to trace a single sample through the network.
Epoch
0
Train Loss
—
Train Acc
—
Val Acc
—
Params
0

Train the network to see learned filters.

Sample 0 / 0

Train the network first, then browse inference results.