Name a goal. Get the map. Walk it.
KFlow turns anything you want to learn into a map of concepts in the order they build on each other, then teaches each one with lessons you can question, instruments you can run and practice that tells you what stuck.
Understand how neural networks learn
From a goal to mastery in four moves
You bring the question. KFlow brings the structure, one level at a time.
Say what you want to learn
A goal in your own words and one question about what you already know.
Get your map
Concepts arrive in levels, each one built on the ones before it, streamed as they are written.
Learn each concept
Lessons, running instruments, and questions you can ask about any sentence.
Prove it and go further
Practise a whole level, mark concepts learned, then unlock the next level.
Sixteen subjects, all hands-on
Every subject comes with instruments you can run, not just read about.
Physics
Mathematics
Biology
Chemistry
Algorithms
Probability
Economics
Linguistics
Logic
Geography
History
Psychology
Sociology
Civics
Earth science
Astronomy
Lessons built on how memory works
Every lesson follows the same rhythm, drawn from learning science, and this one is live: answer the warm-up, reveal the quiz, or select any sentence.
Warm up first
A question that pulls up what you already know before anything new arrives.Stop and explain
Short prompts that make you say the idea back in your own words.Questions at every depth
Bloom-levelled questions, from remembering a fact to applying it to something new.Ask about any sentence
Select text to ask a question or keep a note; the answer links to concepts in your other paths.Gradient descent
Before You Begin...
You are standing on a foggy hillside and want to reach the valley floor. You can only feel the ground under your feet. How would you decide which way to step?
Building On What You Know
You already know that a derivative measures slope. Gradient descent turns that slope into a direction to move, and backpropagation, next, is how a network finds the slope for every weight at once.
Rolling downhill on purpose
A model's loss is a single number that says how wrong it is. Picture every possible setting of the model's weights as a point on a landscape, and the loss as the height at that point. Learning means finding a low spot.
Gradient descent does exactly what you would do in the fog: feel the slope where you stand, then take a small step downhill.
weights = weights - learning_rate * gradient(loss, weights)The gradient points uphill, so we step against it. The learning rate decides the step size: too small and training crawls, too large and you overshoot the valley and bounce around it.
In one sentence, why do we subtract the gradient instead of adding it?
Your loss goes down for a few steps, then starts jumping up and down wildly. What would you change first, and why?
Don't just read it. Run it.
Lessons embed live instruments: change a value, press play and watch the idea behave. Here are eight of the more than 130 across 16 subjects.
A pendulum: change the length and gravity, and watch the period respond.
A bob swings from a pivot under tension and gravity.
ready — bob at -75.1° from vertical; press Step or Run to swing
One map of everything you know
Every path you start joins the same map. When two paths share an idea, KFlow draws the link, so what you learned for one goal speeds up the next.
A companion that notices
It reads your progress and suggests the next concept, a review or a new path, and you decide what to accept.Learn alongside others
See who is learning the same concept right now and open an anonymous conversation with them.Bring what you have
Paste notes or drop a Markdown file, and KFlow turns it into a path you can keep building.KFlow in numbers
130+
Interactive instruments
16
Subjects
6
Bloom levels of questions
3
Languages: English, Arabic, Slovenian
What do you want to understand next?
Tell KFlow your goal and get your first level of concepts in under a minute. Free to start.