Module 1: What a Large Language Model Is
1 min read·13 Oct 2026
What you need first
- Basic Python: lists, dictionaries and functions
- A Google account for Colab, and a free Groq or Gemini key (course-guide/02)
- No earlier module: the course and the
supportdesk/package start here
What you will build
The supportdesk/ package with the shared data, one client for three providers, a first triage call scored with an interval, TinyLM trained and saved, and a model selection.
Lessons
| File | Title | Minutes | Depth tags |
|---|---|---|---|
01-next-token-prediction.md | Lesson 1: A model that only predicts the next token | 50 | Essential, Recommended |
02-what-pretraining-learned.md | Lesson 2: What pretraining gave the model, and what it costs | 40 | Essential, Recommended, Optional |
03-from-base-model-to-assistant.md | Lesson 3: How a text predictor becomes an assistant | 45 | Essential, Recommended |
04-capabilities-and-limits-measured.md | Lesson 4: Measuring what a model can and cannot do | 45 | Essential, Recommended |
05-the-model-landscape.md | Lesson 5: The model landscape | 45 | Essential, Recommended |
Extras
06-module-lab.md: Lab: Start the support desk package (60 minutes): the package, a scored triage baseline, TinyLM saved, nine keyless tests and the first commit.07-incident-file.md: Incident: A reading list of books that do not exist (30 minutes): a May 2025 newspaper list of invented books, replayed with SmolLM2 and fixed by writing references in code.