Package ai.djl.training.tracker

Contains classes for having a gradually changing hyper-parameter.

See: Description

Interface Summary

Interface	Description
Tracker	A Tracker represents a hyper-parameter that changes gradually through the training process.

Class Summary

Class	Description
CosineTracker	CosineTracker is an implementation of Tracker which is updated by taking sections of a cosine curve to smoothly reduce learning rate until a specified step and base learning rate.
CosineTracker.Builder	The Builder to construct an CosineTracker object.
FactorTracker	FactorTracker is an implementation of Tracker which is updated by a multiplicative factor.
FactorTracker.Builder	The Builder to construct an FactorTracker object.
LinearTracker	FactorTracker is an implementation of Tracker which is updated by a constant factor.
LinearTracker.Builder	The Builder to construct an LinearTracker object.
MultiFactorTracker	MultiFactorTracker is an implementation of Tracker which returns piecewise constant values for fixed numbers of steps.
MultiFactorTracker.Builder	The Builder to construct an MultiFactorTracker object.
WarmUpTracker	A WarmUpTracker applies a simple warm-up before executing a main Tracker.
WarmUpTracker.Builder	The Builder to construct a WarmUpTracker.

Enum Summary

Enum	Description
WarmUpTracker.Mode	An enum that enumerates the types of warm-up modes for a WarmUpTracker.

Package ai.djl.training.tracker Description

Contains classes for having a gradually changing hyper-parameter.

It contains a main interface Tracker and various options that extend it.