2001 · Master's thesis
A neural network, watched while it learns.
Self-organizing networks with every step of the learning drawn on screen, so you can see the map form instead of trusting the numbers. Rebuilt for the browser, with verification notes. The 2001 look was kept on purpose.
Open it
Runs in the browser. Nothing to install.
What forms inside a network while it learns, and can you see it instead of inferring it from the loss? In 2001 that was a thesis question. Today it is a field: mechanistic interpretability.