Available draft · two ninety-minute units · construction edition
Attempt the student work before consulting the solutions. Untouched exercises intentionally report NEEDS_WORK.
| Session | Notebook | Matching text | Purpose |
|---|---|---|---|
| A · 90 minutes | A: agent evaluation | Markdown | Construct, connect and explain |
| B · 90 minutes | B: evaluation statistics | Markdown | Derive, check coverage and transfer |
Scroll horizontally to view columns.
Each notebook includes its own setup, first-principles introductions and supplied teaching runtime. Open either notebook in Google Colab with its badge, or use a local Python 3.12+ Jupyter kernel; Unit A also needs Pydantic 2, and Unit B only the standard library. No repository checkout, previous notebook kernel or live account is needed. Unit A closes with an extension that puts a language model behind the chapter's own tools: a recorded transcript runs everywhere, and an OpenAI key in Colab's Secrets pane switches the same loop to a live gpt-5.1 run. Unit B puts error bars on an evaluation: Wilson intervals checked by exact coverage, and McNemar's paired test. Unit B can use an explicitly selected successful Unit A handoff; an invalid selected file refuses rather than substituting an answer.
Ninety minutes is the planned work allowance per unit, excluding installation. Actual completion time and understanding require classroom observation.
Keep building with Prof Rod
Found this material through a colleague, classroom or shared download? Get the complete book at profrod.ai/book and join the Prof Rod learner community. Bring one result, one question or one failure you learned from. Share this resource with another learner and keep its source links with it so they can find the full course and future updates.
