Deep Learning
CS 4715/5715 · University of Idaho · Fall 2026 · Mon/Wed
This is the course site for Deep Learning: an implementation-first path from backpropagation to Transformers, large language models, and diffusion models. You will build an autograd engine, train real networks in PyTorch, implement a Transformer from scratch, fine-tune with LoRA, and finish with a semester project you present on video.
📋 Syllabus & schedule · 🖥 Slides · 📖 Readings · 📓 Notebooks · 🧪 Project
The book — course reading chapters
Textbook-style chapters written for this course. Each chapter section includes worked, runnable code examples, historical context, common pitfalls, and pointers back to the sessions and homework it supports.
Part I — Foundations (Sessions S01–S08)
From linear models to trained multilayer perceptrons: backpropagation, autograd engines, optimizers, regularization, and the PyTorch toolchain. → Chapter 1
Part II — Convolutional Networks & Training Practice (S09–S12)
Convolution, ResNets, transfer learning, and the practical craft of training and debugging deep networks. → Chapter 2
Part III — Sequences, Transformers & Large Language Models (S13–S22)
From RNNs to attention to the full Transformer; pretraining, fine-tuning and LoRA, decoding and prompting, alignment, and multimodal models. → Chapter 3
Part IV — Generative Models, Efficiency & Evaluation (S23–S27)
VAEs, GANs, and diffusion models; training and serving efficiently; evaluating models honestly and safely. → Chapter 4
Milestones at a glance
| When | Event |
|---|---|
| S04 → S08 | HW1 — micro-autograd |
| S09 | Quiz 1 (Module 1) |
| S11 | Project guide released |
| S12 → S18 | HW2 — CNNs & transfer learning |
| S13 | Quiz 2 (Module 2) |
| week of S13 | Project pre-proposal due |
| S15 | Not held this term; the S01–S13 review guide is for self-study |
| S18 → S22 | HW3 — Transformer from scratch |
| S23 | Quiz 3 (Module 3) |
| week of S23 | Project proposal due |
| S26 → S29 | HW4 — evaluation & efficiency |
| S27 | Quiz 4 (Module 4) |
| week of S27 | Project final report due |
| Sunday before S28 | Project recorded presentation (video) due |
| S28 & S29 | Code defenses, for students the instructor calls |
Every date is on the Canvas calendar, which is authoritative; this site is ordered by session so that it stays correct when the calendar shifts.
The full session-by-session schedule, grading, and policies are in the syllabus.