Created by David Bachman with GPT 5.5. To learn more about David see https://pzacad.pitzer.edu/~dbachman/, and subscribe to his AI podcast Entropy Bonus at https://profbachman.substack.com/.
Backprop Builder is a browser-based teaching app for small tensor backpropagation, with an editable computation graph, forward and backward pass controls, loss inspection, gradient visualization, and parameter updates.
Use the graph canvas to build or adjust a small computation graph, edit input and parameter values in the nodes, then step through the forward pass, loss, backward pass, and update controls to see how gradients drive learning. Use save and import to continue a workspace later.