PROJECT FILM / ZONGO MAQUTU

MNIST digits through a 3D neural network

Name: MNIST digits through a 3D neural network
Uploaded: 2026-10-01T13:52:59+02:00
Duration: PT35.13S
Description: Watch real handwritten inputs, hidden-layer activations and output predictions in a freshly trained browser neural network. This is a recording of a local application run. It illustrates individual predictions and network activity; it is not a benchmark of overall model accuracy.

Watch real handwritten inputs, hidden-layer activations and output predictions in a freshly trained browser neural network.

Fresh local run · 35 seconds · actual input digits and predictions, with an orbiting camera. · Silent demonstration.

What the film shows

The input layer displays a handwritten MNIST digit.
The camera moves around the 784–16–16–10 network so its layers remain visible.
The labeled output and prediction readout show the result of the same activation package.

Behind the demonstration

This is a recording of a local application run. It illustrates individual predictions and network activity; it is not a benchmark of overall model accuracy.

Built and documented by Zongo Maqutu. Read the Field Note for the implementation, architecture and limitations.

Read Field Note