Explain it back: machine learning
A free Claude skill that checks your understanding of how machine learning works.
What it does
You explain it in your own words; it tells you what you got right, names what's missing, and asks one question per gap until you've got all four steps: Loss, Direction, Move, Repeat.
An example
From the skill's own README:
- You type
Explain it back: a model learns by looking at lots of data and getting it right eventually.
- It replies, roughly
What you got right: the model learns from data, and it gets better over time.
What's missing: how it knows how wrong it is; how it decides what to change.
1. Before a model can improve, what does it need to measure about its guess?
2. Once it knows how wrong it is, how does it know which way to change?
How it works
It compares your words with the four steps, marks each one as got it, partly, missing or wrong, says what you got right first, then asks at most three questions at a time, most important first. It loops until the four steps are there and nothing you've said is wrong.
- Uses only facts from the post and its sources (Google's Machine Learning Crash Course and glossary)
- If you're stuck after two tries on the same gap, it explains that one idea simply, then asks you to say it back
Get it

A free tool, published under the MIT licence by Elle the Engineer.
Next project