Asked about Regularisation and validation, hands on
What is the difference between L1 and L2 regularisation?
An answer that works
L2 penalises squared weights and shrinks them all smoothly. L1 penalises absolute weights and drives many to exactly zero, which also performs feature selection.
Reading an answer is not being able to give one. The section that teaches this is Regularisation and validation, hands on, 5 hours of Overfitting & regularisation.
Learn this properly