Class LambdaLRScheduler

Namespace

AiDotNet.LearningRateSchedulers

Assembly

AiDotNet.dll

Sets the learning rate using a user-defined lambda function.

public class LambdaLRScheduler : LearningRateSchedulerBase, ILearningRateScheduler

Inheritance

object

LearningRateSchedulerBase

LambdaLRScheduler

Implements

ILearningRateScheduler

Inherited Members

LearningRateSchedulerBase._baseLearningRate

LearningRateSchedulerBase._currentLearningRate

LearningRateSchedulerBase._currentStep

LearningRateSchedulerBase._minLearningRate

LearningRateSchedulerBase.CurrentLearningRate

LearningRateSchedulerBase.BaseLearningRate

LearningRateSchedulerBase.CurrentStep

LearningRateSchedulerBase.Step()

LearningRateSchedulerBase.GetLearningRateAtStep(int)

LearningRateSchedulerBase.Reset()

LearningRateSchedulerBase.LoadState(Dictionary<string, object>)

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Examples

// Custom schedule: lr = base_lr * (0.95 ^ epoch)
var scheduler = new LambdaLRScheduler(
    baseLearningRate: 0.1,
    lrLambda: step => Math.Pow(0.95, step)
);

// Warmup for 10 steps, then constant
var warmupScheduler = new LambdaLRScheduler(
    baseLearningRate: 0.001,
    lrLambda: step => step < 10 ? (step + 1) / 10.0 : 1.0
);

Remarks

LambdaLR provides maximum flexibility by allowing you to define any learning rate schedule as a function of the current step. The lambda function takes the step number and returns a multiplier that is applied to the base learning rate.

For Beginners: This scheduler lets you define your own custom learning rate schedule using a function. The function receives the current step number and returns a value that gets multiplied with the initial learning rate. For example, returning 0.5 would give half the initial learning rate. This is useful when you want a schedule that doesn't fit any of the standard patterns.

Constructors

LambdaLRScheduler(double, Func<int, double>, double)

Initializes a new instance of the LambdaLRScheduler class.

public LambdaLRScheduler(double baseLearningRate, Func<int, double> lrLambda, double minLearningRate = 0)

Parameters

baseLearningRate double

The initial learning rate.

lrLambda Func<int, double>

A function that takes the step number and returns a multiplier for the base learning rate.

minLearningRate double

Minimum learning rate floor. Default: 0

Methods

ComputeLearningRate(int)

Computes the learning rate for a given step.

protected override double ComputeLearningRate(int step)

Parameters

step int

The step number.

Returns

double

The computed learning rate.

GetState()

Gets the scheduler state for serialization/checkpointing.

public override Dictionary<string, object> GetState()

Returns

Dictionary<string, object>

A dictionary containing the scheduler state.