Easy data structure and algorithm practice puzzle.
Given a list of activations and a mask of 0s and 1s, return the activations with every masked-out spot set to 0. Multiply each value by its mask. Dropout randomly switches off neurons during training, so no neuron can lean too hard on any other. It's only for training. Leave it on when predicting and you'll get different answers every time.
Solve it three ways: snap the jigsaw pieces together, fill in the blanks, or type it from scratch against the tests. This one is free.
Every line of the solution comes with a note explaining why it is there, and a step by step replay shows the code running. At level 3, where you type it yourself, you can ask for an AI mock interview on this problem that ends with a hiring manager's scorecard.
Then grow it into a real project with the Software Engineer button: this puzzle becomes the Sample Ranking Service, a pipeline you can run and download. See every Software Engineer project.
Browse all 576 puzzles