promptdojo_
Checkpoint

One last thing before we move on. Same surface as a write step — but the lesson doesn't complete until this passes.

Final drill for this lesson. Write classify_framings(framings) that takes a list of dicts shaped like:

{"statement": "...", "correct": True or False}

and returns a TUPLE of two things:

  1. correct_count — an integer, how many of the framings are marked correct: True.
  2. wrong_statements — a list of strings, the statement text from every framing marked correct: False, in the order they appeared in the input.

Five framings run. Expected output:

correct: 3
wrong:
  - When I correct an LLM mid-conversation, the model updates its weights so the next user sees the corrected answer.
  - LLMs are essentially search engines that retrieve the most relevant answer from their training corpus.

The point of this drill: you can now read these five statements and FEEL which ones are right. That feeling is the mental model this lesson was designed to install. The function just makes the classification explicit.

this step needs the editor

on desktop today; in the app (coming soon). save your spot and we'll bring you back here when you're ready.

open this same url on a laptop to keep going today.