AIDE500 · Programming for AI

AIDE500 exam prep

A 12-lesson crash course on Labs 1–10, the 47-question mock exam solved and explained, and a cheat sheet for the last hour. Your progress stays in this browser.

Lesson 0 · Overview

How the ten labs fit together

Every lab builds one piece of a single pipeline: images go in, a network guesses, and the guesses are used to improve the network. See where each piece fits first.

Dataset + transformsload and prepare each imageL8 · L10 DataLoaderbatches of 16, shuffledL8 Network (nn.Module)images → 10 class scoresL6 · L7 Losshow wrong? one numberL9 loss.backward()autograd fills .gradL5 optimizer.step()the weights moveL5 · L9 next batch afterwards: save & test (L9), reuse nets (L11) all data is tensors (L4) ≈ NumPy arrays (L3) run in Jupyter (L1) · plot with matplotlib (L2)
The whole course on one page. The loop (Dataset → … → step → next batch) is what "training" means. The two bars at the bottom are the tools everything else runs on.
LessonLabYou learn
1Lab 1Jupyter: cells, modes, shortcuts
2Lab 2matplotlib: line plots, legend, subplots
3Lab 3NumPy: arrays, shape, ranges, axis, broadcasting
4Lab 4Tensors: create, join, multiply, share memory
5Lab 5 (+6)Autograd: gradients, backward, step, freezing
6Labs 6–7Layers: Linear, Conv2d, pooling, activations
7Lab 7Building a network class and tracking its shapes
8Lab 8Datasets, transforms, DataLoaders
9Lab 9Training and testing the CIFAR-10 classifier
10Lab 10Data augmentation
11Lab 6 + scriptTransfer learning, train/eval mode

Each lesson ends with quick-check cards: tap one to see the answer. The chips under each title jump to the mock exam questions that test that lesson.