Session 2: Colab Onboarding & Your First LLM Call

Day 1 · Introduction & First Experience (1-2)

Learning objectives

By the end of this session you will be able to:

  • Sign in to Google Colab, run a cell (Shift+Enter), and read a Python error message well enough to fix a typo.
  • Store a value in a variable, and send a prompt to a language model with ai.generate_text(...).
  • Build a prompt from a variable with an f-string, and recognise the .format() template form you will edit from Session 7 on.
  • Recognise Python’s core data types — str, list, dict — in what the model hands back.
  • Build the {id, text, label} record the rest of the course uses, with a label you decided yourself, and add the model’s answer beside it.
  • Index and slice into a list of records to compare the two answers.

Agenda

This is a guided, run-along session using Part A of the Day 1 notebook, in eight short steps. The model comes first and stays first: you call it, then learn just enough Python to read and score what it gives back.

  1. Run a cell — sign in, cells and the runtime, Shift+Enter.
  2. Read an error — run a cell that fails on purpose, and read the last line.
  3. Variables — store a value under a name with =.
  4. Your first LLM call — the setup cell (one import), then five prompts: the task itself, asking for a format, the same prompt twice, and one the model gets wrong.
  5. f-strings — drop a sentence into a prompt with f"...{sentence}", then write a .format() template of your own.
  6. What kind of value is the answer? — type(), and the three types you use all week.
  7. Put the answer in a record — decide a sentence’s level yourself, then ask the model the same question and store both in one {id, text, label} record.
  8. Indexing and slicing — items[2]["label"], items[:2], and comparing the two answers with ==.

Control flow (for, if, elif, functions), counting, and text segmentation come next, in Session 3.

Reading

No new reading for this session — the Day 1 reading (Abdurahman et al., 2025, skim) is listed in Session 1 and on the Readings page.

Slides & Colab