What is learning rate decay?

Updated May 16, 2026

Short answer

Learning rate decay gradually reduces learning rate over time.

Deep explanation

Decay strategies help stabilize training by taking large steps early and smaller steps later, improving convergence to minima.

Real-world example

Deep neural network training stabilization.

Common mistakes

  • Reducing learning rate too aggressively.

Follow-up questions

  • What are decay strategies?
  • Why decay helps?

More Gradient Descent interview questions

View all →