PYTHON · NUMPY
Curiosity: seeking the reflected wave

Two learners, one budget. The curious one trains on the experiences it predicts worst; the uniform one trains on random draws from the same pool. Watch when curiosity wins — and run the honest catch. The third cell asks which *kind* of surprise: prediction error (a proxy) or information gain (OIST's KL term) — and shows they come apart under observation noise.

You read

the arrays and values already in scope

You change

the code you write in each cell

Fixed

the dataset and the checks that grade you