SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-l2-regularization-squares

What does L2 regularization penalize?

The sum of the squares of the weights.

Start playing free

  1. The sum of the absolute weights
  2. The count of nonzero weights
  3. The sum of the squared weights ✔

Why: L2 regularization adds a penalty for big weights to the loss. It pulls large weights toward zero, but not all the way, which keeps the model simpler and less likely to overfit.

Huh, didn't know that: Training then minimizes loss plus complexity, with a regularization rate (lambda) setting the balance.

The dad joke
L2 regularization is a diet for weights. Big ones get squared away.
Like this card?
Play the game

References

Answer: The sum of the squared weights. L2 regularization adds a penalty for big weights to the loss. It pulls large weights toward zero, but not all the way, which keeps the model simpler and less likely to overfit.

  1. Google Machine Learning Glossary: L2 regularization developers.google.com/machine-learning/glossary
    A type of regularization that penalizes weights in proportion to the sum of the squares of the weights.
    Checked 2026-10-10.
  2. Google ML Crash Course: L2 regularization developers.google.com/machine-learning/crash-course/overfitting/regularization
    higher values leading to simpler models and lower values increasing the risk of overfitting.
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWhich regularization can push useless features' weights to exactly zero?Artificial IntelligenceIn gradient descent, which way does each step move the weights?Artificial IntelligenceWhat did AlexNet's authors use to reduce overfitting in its fully connected layers?

That's stuff everyone should know.