Class NAdam
java.lang.Object
ai.beestream.jitnn.nn.optim.Optimizer
ai.beestream.jitnn.nn.optim.NAdam
NAdam (Dozat), PyTorch's semantics: momentum decay psi (default 0.004), L2 weight decay.
-
Field Summary
-
Constructor Summary
Constructors -
Method Summary
Modifier and TypeMethodDescriptionprotected voidRestores whatOptimizer.saveScalars(StateDict)saved.protected voidScalar state beyond the step count and learning rate, for checkpoints.state()Per-parameter state by slot name, each array indexed like the parameters (null where none).voidstep()Records one update of every parameter that has a gradient.Methods inherited from class Optimizer
clipGradNorm, learningRate, loadStateDict, setLearningRate, stateDict, takeStats, zeroGrad
-
Constructor Details
-
NAdam
-
-
Method Details
-
step
-
state
-
saveScalars
Description copied from class:OptimizerScalar state beyond the step count and learning rate, for checkpoints.- Overrides:
saveScalarsin classOptimizer
-
loadScalars
Description copied from class:OptimizerRestores whatOptimizer.saveScalars(StateDict)saved.- Overrides:
loadScalarsin classOptimizer
-