🌏 中文版
Video status: Videos included. Source details
Lectures 19–25 move from rational decisions and value of perfect information to decision trees, linear regression, Naive Bayes, neural networks, language models, and fine-tuning. Project 5 implements the second half in PyTorch through nonlinear regression, digit and language classification, CNNs, attention, and an optional character-GPT.
This is not a replacement for a full deep-learning course
P5 introduces the modern ML workflow inside a broad AI course: model definition, forward pass, loss, optimizer, batches, and training loop. Its breadth does not replace a dedicated course in optimization, representation learning, or large models. Treat each task as an interface exercise instead of trying to acquire all underlying theory at once.
Make the small model explainable first
In regression, explain input and output shapes, falling loss, and held-out behavior before advancing. Digit classification adds multiclass loss; language identification adds sequences; the CNN task introduces spatial inductive bias; and attention requires tracking query, key, and value tensor relationships.
The official P5 specification warns that NumPy 2.0 can cause compatibility failures and suggests 1.24.3 or another version below 2.0 when those errors appear. Treat this as environment reproduction, not a reason to hide the mismatch with a model-code workaround.
For every task, record a baseline, loss curve, and one failure case. The local autograder can test interfaces and thresholds, but it cannot explain why a model fails. That example is the analysis independent study must add.
Series navigation: Previous: Bayes nets and Ghostbusters | Next: Completion route
Course video sources
The official Spring 2026 schedule and the official YouTube playlist ([CS188 SP26] Live Lectures, 28 videos) were checked live on 2026-10-10; the lecture recording embedded here is listed there as a “Recording”. No unverified timestamp is supplied.
Original videos: CS188 Spring 2026 Lecture 19: Rational Decisions
Course and recording entries:
Checked: 2026-10-10.
Content check: verified against the video transcript (2026-10-10): sampled check (only the first ~5,000 characters of the transcript were read): the embedded video is Lecture 19, on Rational Decisions; the transcript covers rational agents maximizing utility, why utilities can be numbers (using minimax's invariance under monotonic transformations), why expected utility is used, and previews RLHF and alignment, matching this article's link from rational decisions to ML. The remaining sections of this article cover other lectures in the unit and the project, not this video.
Update Log
- 2026-10-10: Added explicit video status and checked recording sources and access notes.
- 2026-10-10: Rechecked video status. The official Spring 2026 schedule and YouTube playlist were checked live and list the embedded lecture recording, so the status is now Videos included.
- 2026-10-10: Checked the video content against its transcript. Confirmed the embedded video is Lecture 19 (Rational Decisions) and fits the article scope; no video claims needed correcting.
References
Loading...