# CS221 lecture companion sources

The lecture path covers the 19 lecture entries supplied for Stanford CS221 Autumn 2025. It sits alongside the existing foundations curriculum.

## Original notes and code

- [Official lecture repository](https://github.com/stanford-cs221/autumn2025-lectures)
- [Official course schedule](https://stanford-cs221.github.io/autumn2025/)
- [Original executable lecture viewer](https://stanford-cs221.github.io/autumn2025-lectures/)

The Python sources and local diagrams were retrieved on September 17, 2026. Their MIT license and Stanford copyright notice are included in `stanford/LICENSE`. All Python source lines are retained. The reading displays computational examples beside the related transcript explanation, with lecture-presentation commands available in expandable original-source views. Original programs may require NumPy, PyTorch, edtrace, Graphviz, and other dependencies. Follow the original repository's setup instructions for those programs.

The language-model slides name Ken Liu. The Economics / AI Supply Chains slides name Rishi Bommasani. The Study Lab's economics and society numerical experiments are explicitly hypothetical, not empirical company or population data.

## Corrected videos

- Lecture 13, Gibbs sampling: https://www.youtube.com/watch?v=Dk7Kqqehzjk
- Lecture 14, Bayesian networks and learning: https://www.youtube.com/watch?v=4d9V6Sxa6gU
- Lecture 15, Logic I: https://www.youtube.com/watch?v=Q7V13XriJEc

The supplied lecture-13 link pointed to Logic I, and the supplied lecture-14 link pointed to Gibbs sampling. Titles and caption tracks were verified against Stanford Online's videos. The Games I video ID uses a literal underscore, with the Markdown escaping removed. Detailed source records are in `video-manifest.json`.

The course schedule and video call lecture 19 “AI Supply Chains”; the repository calls the same December 1 lecture “Economics of AI.” Both names appear in the companion.

## Full user-supplied transcripts

The complete lecture chapters use the transcript text supplied by the user, including their separate correction for Lecture 12. The originally uploaded Lecture 12 repeated Games II; that information remains in Lecture 11, and Lecture 12 now contains the complete Bayesian-networks lecture. The text is cleaned into paragraphs with literal transcription corrections, headings and readable mathematical notation. Substantive examples, questions, qualifications, discussions and course information are retained.

Each paragraph is mapped to its exact original input span for coverage verification. No excerpt selection or summary replaces the transcript. English captions previously retrieved using yt-dlp were used for short excerpts in the earlier companion; those are no longer the reading page. The original videos remain linked in the source footer. Embedded players, transcript-access instructions and inline quizzes are removed from the lecture reading.

## Runnable examples and progress

All 61 Study Lab Python examples were executed under the embedded Pyodide 0.27.7 runtime, and their output was compared with the displayed reference output. Code runs locally in a fresh worker with a Stop control and a 15-second execution limit. Edits stay in the current tab; exercise progress syncs through the existing private account progress service. Hint use and revealed answers do not grant independent-practice credit.