Course links
| Course | Links | Role |
|---|---|---|
| UMich EECS 498-007 / 598-005, Fall 2019 — Justin Johnson, Deep Learning for Computer Vision | Course · Schedule, slides, and readings | Main course: networks, training, convolutional and recurrent networks, attention, hardware, detection, and segmentation. Its assignments implement each component from scratch in PyTorch. |
| UNIGE 14x050 — François Fleuret, Deep Learning | Course page with slides, handouts, and screencasts | Supplement: short PyTorch-centered sections on the same topics, with autoencoders, attention, and practical sessions. |
| Stanford CS224W, 2021 — Jure Leskovec, Machine Learning with Graphs | Course · Lecture playlist | Optional chapter 13: graph neural networks (lectures 6 to 9). |
Three practical guides complement the courses:
| Guide | Link | Role |
|---|---|---|
| Andrej Karpathy, A Recipe for Training Neural Networks (2019) | Blog post | The workflow of chapter 12: inspect the data, build baselines, overfit, then regularize and tune. |
| Godbole, Dahl, Gilmer, Shallue, and Nado, Deep Learning Tuning Playbook (2023) | Repository | Hyperparameter tuning as a scientific process (chapter 12). |
| Hugging Face, The Ultra-Scale Playbook (2025), and Google DeepMind, How to Scale Your Model (2025) | Ultra-Scale Playbook · How to Scale Your Model | Memory, parallelism, and hardware for training at scale (chapter 11). |
The reading plan maps individual lectures, slides, and readings to the study sequence. Video links arranges the lecture recordings by chapter, and the DL overview maps the reading-plan topics to the chapters.