Uses of Interface
org.nd4j.autodiff.samediff.optimize.Optimizer

Packages that use Optimizer

Package	Description
org.nd4j.autodiff.samediff.optimize
org.nd4j.autodiff.samediff.optimize.debug
org.nd4j.autodiff.samediff.optimize.optimizations

Uses of Optimizer in org.nd4j.autodiff.samediff.optimize

Methods in org.nd4j.autodiff.samediff.optimize that return types with arguments of type Optimizer

Modifier and Type	Method and Description
List<Optimizer>	OptimizerSet.getOptimizers()

Uses of Optimizer in org.nd4j.autodiff.samediff.optimize.debug

Methods in org.nd4j.autodiff.samediff.optimize.debug with parameters of type Optimizer

Modifier and Type	Method and Description
void	OptimizationDebugger.afterOptimizationsCheck(SameDiff sd, SameDiffOp op, Optimizer o, boolean wasApplied)
void	OptimizationDebugger.beforeOptimizationCheck(SameDiff sd, SameDiffOp op, Optimizer o)

Uses of Optimizer in org.nd4j.autodiff.samediff.optimize.optimizations

Classes in org.nd4j.autodiff.samediff.optimize.optimizations that implement Optimizer

Modifier and Type	Class and Description
static class	ConstantFunctionOptimizations.FoldConstantFunctions
static class	CuDNNFunctionOptimizations.CudnnConv2dNCHWtoNHWCConversion https://docs.nvidia.com/deeplearning/sdk/dl-performance-guide/index.html#tensor-layout For tensor cores: we want NHWC layout: Section 7.3.1 "Layout choice has an effect on performance, as convolutions implemented for Tensor Cores require NHWC layout and are fastest when input tensors are laid out in NHWC." "To maximize performance, we recommend using NHWC tensor layout." As for weights format: cuDNN docs are vague - but TF uses NCHW+OIHW or NHWC+OHWI
static class	IdentityFunctionOptimizations.RemoveIdentityOps Remove identity(x)
static class	IdentityFunctionOptimizations.RemoveIdentityPermute Remove permute(0,1,2,...,rank-1) as this is a no-op
static class	ShapeFunctionOptimizations.FuseChainedConcatOps Fuse [concat(concat(concat(x,y,dim=D), z, dim=D), a, dim=D)] into a single concat op, concat(x,y,z,a, dim=D) As long as the intermediate outputs aren't needed elsewhere
static class	ShapeFunctionOptimizations.FuseChainedPermutes Fuse [permute1 -> permute2 -> ...
static class	ShapeFunctionOptimizations.FuseChainedReshapes Fuse [reshape1 -> reshape2 -> ...
static class	UnusedFunctionOptimizations.RemoveUnusedConstants

Methods in org.nd4j.autodiff.samediff.optimize.optimizations that return types with arguments of type Optimizer

Modifier and Type	Method and Description
List<Optimizer>	BaseOptimizerSet.getOptimizers()