Skip to content
AI360Xpert
Core ML
Visual explainer

Learning Rate Scheduling, Visually

Five pictures: How to adjust the learning rate during training to escape plateaus, avoid wild oscillations, and smoothly settle into the optimal minimum.

A static learning rate is too slow at the start and too volatile at the end.
A static learning rate is too slow at the start and too volatile at the end.

A single, fixed learning rate is a difficult compromise. If it is kept small, the model crawls across initial flat plateaus and takes an eternity to learn. If it is large, it bounces chaotically around the steep walls of the minimum at the end, completely unable to settle.

Step Decay

Step decay drops the learning rate by a fixed factor at regular epoch intervals.
Step decay drops the learning rate by a fixed factor at regular epoch intervals.

Step decay simply reduces the learning rate by a fixed multiplier at predefined training milestones. This tiered approach allows aggressive, fast progress early on, followed by smaller, fine-grained steps that precisely lock into the deepest part of the minimum.

Cosine Decay

Cosine decay gracefully anneals the learning rate to zero following a cosine curve.
Cosine decay gracefully anneals the learning rate to zero following a cosine curve.

Instead of relying on harsh discrete jumps, cosine decay smoothly and continuously anneals the learning rate down to zero. This continuous mathematical descent completely avoids sudden shocks to the network weights, resulting in a significantly more stable training trajectory.

Linear Warmup

Warmup gradually scales the learning rate up from zero to stabilize early training.
Warmup gradually scales the learning rate up from zero to stabilize early training.

At the very beginning of training, entirely random weights generate massive, noisy gradients. A linear warmup phase starts the learning rate near zero and gradually increases it to peak capacity, preventing these early wild updates from destroying the network.

Where It Breaks

Decaying the learning rate too early leaves the model permanently stranded in poor local minima.
Decaying the learning rate too early leaves the model permanently stranded in poor local minima.

Decaying the learning rate is fundamentally a one-way trip. If the learning rate drops too drastically and too prematurely, the optimizer entirely loses the momentum and energy needed to escape small local dips, permanently leaving it stranded far from the global minimum.

The Quick Version

  • A static step size is too slow early on and too chaotic later.
  • Step decay drops the rate at fixed epoch milestones like literal stairs.
  • Cosine decay continuously and smoothly anneals the learning rate down to zero.
  • Linear warmup protects fragile initial weights by ramping up the rate slowly.
  • The honest failure mode: Dropping the rate prematurely traps the model permanently.

What to Read Next