Class ConstantLRScheduler
- Namespace
- AiDotNet.LearningRateSchedulers
- Assembly
- AiDotNet.dll
Maintains a constant learning rate throughout training.
public class ConstantLRScheduler : LearningRateSchedulerBase, ILearningRateScheduler
- Inheritance
-
ConstantLRScheduler
- Implements
- Inherited Members
Examples
var scheduler = new ConstantLRScheduler(baseLearningRate: 0.001);
Remarks
ConstantLR simply returns the same learning rate for every step. While this is the simplest scheduler, it can be useful as a component in composite schedulers or for fine-tuning where you want to keep the learning rate fixed.
For Beginners: This is the simplest scheduler - it just keeps the learning rate the same throughout training. While adaptive schedules often work better, sometimes you want a fixed learning rate, especially for fine-tuning or when the learning rate has already been carefully tuned for your specific problem.
Constructors
ConstantLRScheduler(double)
Initializes a new instance of the ConstantLRScheduler class.
public ConstantLRScheduler(double baseLearningRate)
Parameters
baseLearningRatedoubleThe constant learning rate to maintain.
Methods
ComputeLearningRate(int)
Computes the learning rate for a given step.
protected override double ComputeLearningRate(int step)
Parameters
stepintThe step number.
Returns
- double
The computed learning rate.