The build output for an interactive tutorial on reading Python's memory model, generated from a nine-stage exercise collection and rendered with the University of Toronto's MemoryViz.
This page is the bundle inspector, not the tutorial. The front-end has not been built yet. What follows reads the published JSON directly, so the data can be checked and the trace format seen working. The stepper at the bottom is the real artefact: every state it shows was produced by executing the exercise under CPython, not authored by hand.
Ground truth for most items is derived: the build runs each snippet and reads the answer off the trace, so no human authored it. The remainder need a written supplement before they can be machine-graded.
One bar per grader kind. The tag on the right says whether the build computes the expected answer itself or waits on a human.
Pick an exercise and step through it. The highlighted line is the one about to run; the diagram is the state at that moment. Use ← → once the stepper has focus.