case classPolyEpochDecay(power: Double, maxEpochs: Int) extends LearningRateSchedule with Product with Serializable
A learning rate decay policy, where the effective learning rate
follows a polynomial decay, to be zero by the max_epochs.
Calculation: init_lr * (1 - epoch/max_iteration) ^ (power)
power
The coefficient of decay.
maxEpochs
The maximum number of epochs when lr becomes zero.
Linear Supertypes
Serializable, Serializable, Product, Equals, LearningRateSchedule, AnyRef, Any
Ordering
Alphabetic
By Inheritance
Inherited
PolyEpochDecay
Serializable
Serializable
Product
Equals
LearningRateSchedule
AnyRef
Any
Hide All
Show All
Visibility
Public
All
Instance Constructors
newPolyEpochDecay(power: Double, maxEpochs: Int)
power
The coefficient of decay.
maxEpochs
The maximum number of epochs when lr becomes zero.
A learning rate decay policy, where the effective learning rate follows a polynomial decay, to be zero by the max_epochs. Calculation: init_lr * (1 - epoch/max_iteration) ^ (power)
The coefficient of decay.
The maximum number of epochs when lr becomes zero.