What Is a Learning Rate?

What Is a Learning Rate?

The learning rate is a number that controls how big each training step is when updating a model’s weights. It determines how fast or slow the model learns.

Why Learning Rate Matters

• Too high: model becomes unstable or fails to learn.
• Too low: training becomes extremely slow.
• Just right: smooth, efficient learning.

Choosing a Good Learning Rate

• Start small (0.001 is common)
• Use learning-rate schedules
• Monitor loss during training

Learning Rate FAQ

Can learning rate change during training?

Yes—schedules and optimizers often reduce it over time.

What happens if it’s wrong?

Bad accuracy or training failure.