You think you understand it. Kapi finds out if you do.
Explain a machine-learning concept in your own words. Kapi finds the specific gap in your understanding and helps you close it.
- 1. Explain it
Pick a concept and say what you think it means. - 2. Find the gap
Kapi pinpoints what you specifically get wrong. - 3. Close it
Short questions and hints, then check you really got it.
No sign-up needed. Saving your progress is optional.
See what a session looks like
Kapi
Overfitting — pretend I've never heard of it. How would you describe it?
You
It's when the model gets too good at the training data, so it just memorises it instead of learning.
Kapi
You've got the memorising part, and that's the heart of it. But you've not said how you would ever notice it happening — what would you look at to tell a model that memorised from one that learned?
Rough and half-right is what Kapi wants — that is what it reads to find the gap.
Pick a concept that isn't clicking
- Linear regressionFitting a line to data to predict a number.🗂 4 cards
- Gradient descentHow models learn by stepping downhill on a loss.🗂 4 cards
- OverfittingWhen a model memorizes training data instead of learning.🗂 4 cards
- Train/test splitWhy we hold data back to judge a model honestly.🗂 4 cards
- Logistic regressionPredicting probabilities and classes with a sigmoid.🗂 4 cards
- BackpropagationHow neural networks assign blame to each weight.🗂 4 cards