Slides skip the details. Ritsu helps you master them.

Ritsu reads every slide, walks you back through the reasoning each bullet skipped, and turns any concept into personalized, interactive practice in seconds — until you can produce the reasoning yourself, without the deck.

A 20-second product loop: a document becomes a step-by-step learning path in Ritsu.

The missing middle

You're not failing the slide. You're failing what it left out.

A deck keeps the conclusions and drops the reasoning — and you can't re-read your way to something that was never written down.

Slide 14 says therefore Lw=δa\tfrac{\partial L}{\partial w} = \delta \cdot a^{\top} and moves on. It looks finished. It's clean, it's what the professor put on the screen, and it's right there in front of you — so you read it again.

In the lecture, that one word — therefore — took four minutes and a whiteboard. In the deck it takes one bullet, and the four minutes are gone. You're not failing the slide. You're failing what it left out — and you cannot re-read your way to something that was never written down.

There's only one way out: stop staring at the conclusion. Go back and rebuild the step.

The workflow

From slides you can read to reasoning you can produce.

01

Drop in the deck

Your lecture slides. That's the only manual step.

Lecture 9 · Backpropagation01 / 06
the gap
02

/deriverebuild the step between the slides

Slide 13 sets up. Slide 14 concludes. Ritsu walks the space between them — each step with the reason it's licensed, not just the algebra. Then it stops, hands you the pivotal one, and waits.

Lecture 9 · Backpropagation02 / 06
Lw=Lzzw\tfrac{\partial L}{\partial w} = \tfrac{\partial L}{\partial z} \cdot \tfrac{\partial z}{\partial w}chain rule
Lz=δ\tfrac{\partial L}{\partial z} = \deltathe error signal — slide 13
zw=  ?\tfrac{\partial z}{\partial w} = \; ?your step — z = w·a + b. Try it.
  Lw=δa\therefore \; \tfrac{\partial L}{\partial w} = \delta \cdot a^{\top}substitute back
03

/prereqthe thing it assumed you'd seen

Often, “I don't get slide 14” means “I never really got slide 6 of a different lecture.” Ritsu names what's missing — in learning order — before you lose a weekend to the symptom.

Lecture 9 · Backpropagation03 / 06
Lw=Lzzw\tfrac{\partial L}{\partial w} = \tfrac{\partial L}{\partial z} \cdot \tfrac{\partial z}{\partial w}chain rule
Lz=δ\tfrac{\partial L}{\partial z} = \deltathe error signal — slide 13
zw=  ?\tfrac{\partial z}{\partial w} = \; ?your step — z = w·a + b. Try it.
  Lw=δa\therefore \; \tfrac{\partial L}{\partial w} = \delta \cdot a^{\top}substitute back
/prereqChain rule — assumed known, from an earlier lecture
04

/expandthe bullet doing all the work

“Regularisation prevents overfitting” is six words hiding a whole idea. Ritsu opens it exactly where the deck went thin — and nowhere else.

Lecture 9 · Backpropagation04 / 06
Regularisation prevents overfitting/expand

A penalty on large weights keeps the model from memorising the training set — it has to find the pattern instead.

Smaller weights → a smoother function → answers that still hold on data it has never seen.

opened where the deck went thin — and nowhere else

05

Every concept in the deck, mapped

That was one slide. Ritsu did it for the whole deck — every concept, wired in the order they build on, each one scored as you work.

Lecture 9 · Backpropagation · Knowledge Map05 / 06
Score24%
06

Keep it past the forgetting curve

Ritsu brings each concept back right before you'd forget it. Master the deck today, still own it on exam day.

Lecture 9 · Backpropagation · Knowledge Map06 / 06
Score100%

Pedagogy, packaged

Every command packs a proven learning method.

You've got the reasoning back. Six commands take it from “I can follow it” to “I can write it from memory” — each one asking more of you than the last.

Backpropagation
/explain01

Understand

The core idea first, then the layers — built from your deck, not from the internet.

/quiz02

Recall

Retrieval practice: the most proven way to make it stick. Pull it back out, cold.

/flashcard03

Drill

The parts that have to be automatic, turned into cards you'll actually run.

/askme04

Explain back

The Feynman test. Say it in your own words — that's where the gaps show.

/exercise05

Apply

A real problem on the real concept. Knowing it and using it aren't the same.

/write06

Create

Write it out in full. The highest bar there is: if you can write it, it's yours.

That's Bloom's taxonomy, climbed one command at a time. And the science isn't a slogan — it's a flag on the command:

/quiz --bloom="auto"

40+ commands, all built like this.

Knowledge Map

You always know exactly where you stand.

A deck's slide count tells you nothing about how much there is to learn. Ritsu scores every Point of Knowledge as you work — the dim ones are exactly what you haven't got yet, so you never have to guess what to study next.

Lecture 9 · Backpropagation · Knowledge MapScore 68%
masterednot yetreview due
01

Every concept, scored.

Not one vague % for the deck — a mastery score per Point of Knowledge.

02

Dim = what will bite you.

The concepts you haven't got yet, named. That's the list nobody else gives you.

03

The whole deck, one screen.

No more guessing what to study next — the map tells you.

Spaced review

Master it today. Still have it on exam day.

You forget most of what you learn within days — Ebbinghaus measured it in 1885. Ritsu brings each concept back right before it fades, so the work you did today is still there when it counts.

100%50%0%day 01234567reviewreviewreviewWithout review: gone in a week.With Ritsu: every review resets the curve.

Five minutes tonight. Three tomorrow. Two in three days. Built from your deck — no cards to make.

When you're stuck

The moments you actually get stuck — and what to type.

/derive — Ritsu shows you where it came from — each step with the reason it's licensed. Then it hands you one to try, so it's yours and not just read.

Stop re-reading. Start producing.

Free forever · no card