BatchNorm2d (dfx)

BatchNorm2d Type

Namespace: Dfx.Model
Assembly: Dfx.Core.dll
Base Type: Model

Applies Batch Normalization over a 4D input (a mini-batch of 2D inputs with optional additional channel dimension)

The mean and standard-deviation are calculated per-dimension over the mini-batches and \(\gamma\( and \(\beta\) are learnable parameter vectors of size \(C\) (where \(C\) is the input size). By default, the elements of \(\gamma\) are set to 1 and the elements of \(\beta\) are set to 0. The standard-deviation is calculated via the biased estimator, equivalent to dfx.var(input, unbiased=False).

Also by default, during training this layer keeps running estimates of its computed mean and variance, which are then used for normalization during evaluation. The running estimates are kept with a default momentum of 0.1.

If trackRunningStats is set to False, this layer then does not keep running estimates, and batch statistics are instead used during evaluation time as well.

Constructors

Constructor	Description
`BatchNorm2d(numFeatures, ?eps, ?momentum, ?affine, ?trackRunningStats, ?reversible)` Full Usage: `BatchNorm2d(numFeatures, ?eps, ?momentum, ?affine, ?trackRunningStats, ?reversible)` Parameters: numFeatures : `int` ?eps : `double` ?momentum : `Tensor` ?affine : `bool` ?trackRunningStats : `bool` ?reversible : `bool` Returns: `BatchNorm2d`	numFeatures : `int` ?eps : `double` ?momentum : `Tensor` ?affine : `bool` ?trackRunningStats : `bool` ?reversible : `bool` Returns: `BatchNorm2d`

Instance members

Instance member	Description
`this.bias` Full Usage: `this.bias` Returns: `Tensor`	TBD Returns: `Tensor`
`this.mean` Full Usage: `this.mean` Returns: `Tensor`	TBD Returns: `Tensor`
`this.stddev` Full Usage: `this.stddev` Returns: `Tensor`	TBD Returns: `Tensor`
`this.variance` Full Usage: `this.variance` Returns: `Tensor`	TBD Returns: `Tensor`
`this.weight` Full Usage: `this.weight` Returns: `Tensor`	TBD Returns: `Tensor`

Inherited members

Inherited from `ModelBase`

Instance members

Instance member	Description
`this.addBuffer (buffer, name)` Full Usage: `this.addBuffer (buffer, name)` Parameters: buffer : `Parameter` name : `string`	buffer : `Parameter` name : `string`
`this.addBuffer buffers` Full Usage: `this.addBuffer buffers` Parameters: buffers : `(Parameter * string)[]`	buffers : `(Parameter * string)[]`
`this.addBuffer buffers` Full Usage: `this.addBuffer buffers` Parameters: buffers : `Parameter[]`	buffers : `Parameter[]`
`this.addModel (model, name)` Full Usage: `this.addModel (model, name)` Parameters: model : `Model` name : `string`	model : `Model` name : `string`
`this.addModel models` Full Usage: `this.addModel models` Parameters: models : `(Model * string)[]`	models : `(Model * string)[]`
`this.addModel models` Full Usage: `this.addModel models` Parameters: models : `Model[]`	models : `Model[]`
`this.addModel (model, name)` Full Usage: `this.addModel (model, name)` Parameters: model : `ModelBase` name : `string`	model : `ModelBase` name : `string`
`this.addModel models` Full Usage: `this.addModel models` Parameters: models : `(ModelBase * string)[]`	models : `(ModelBase * string)[]`
`this.addModel models` Full Usage: `this.addModel models` Parameters: models : `ModelBase[]`	models : `ModelBase[]`
`this.addParameter (parameter, name)` Full Usage: `this.addParameter (parameter, name)` Parameters: parameter : `Parameter` name : `string`	parameter : `Parameter` name : `string`
`this.addParameter parameters` Full Usage: `this.addParameter parameters` Parameters: parameters : `(Parameter * string)[]`	parameters : `(Parameter * string)[]`
`this.addParameter parameters` Full Usage: `this.addParameter parameters` Parameters: parameters : `Parameter[]`	parameters : `Parameter[]`
`this.clone ()` Full Usage: `this.clone ()` Returns: `ModelBase`	TBD Returns: `ModelBase`
`this.eval ()` Full Usage: `this.eval ()`	TBD
`this.forwardDiff (derivatives, ?nestingTag)` Full Usage: `this.forwardDiff (derivatives, ?nestingTag)` Parameters: derivatives : `ParameterDict` - The derivatives of the parameters ?nestingTag : `uint32` - The level tag for nested differentiation. Defaults to the current global nesting level	Adjust the parameters of the model to initiate a new level of forward-mode automatic differentiation. After this call the current parameters of the model will have attached derivatives for forward mode differentiation. derivatives : `ParameterDict` The derivatives of the parameters ?nestingTag : `uint32` The level tag for nested differentiation. Defaults to the current global nesting level
`this.init f` Full Usage: `this.init f` Parameters: f : `string * Tensor -> Tensor`	TBD f : `string * Tensor -> Tensor`
`this.move (?device, ?dtype, ?backend)` Full Usage: `this.move (?device, ?dtype, ?backend)` Parameters: ?device : `Device` ?dtype : `Dtype` ?backend : `Backend`	Moves the state (parameters and buffers) of the model to the given configuration ?device : `Device` ?dtype : `Dtype` ?backend : `Backend`
`this.noDiff ()` Full Usage: `this.noDiff ()`	TBD
`this.reverseDiff ?nestingTag` Full Usage: `this.reverseDiff ?nestingTag` Parameters: ?nestingTag : `uint32` - The level tag for nested differentiation. Defaults to the current global nesting level	Adjust the parameters of the model to initiate a new level of reverse-mode automatic differentiation. After this call the current parameters of the model will support reverse-mode differentiation. After the completion of the corresponding `reverse` operation, the computed derivatives will be available. ?nestingTag : `uint32` The level tag for nested differentiation. Defaults to the current global nesting level
`this.summary ()` Full Usage: `this.summary ()` Returns: `string`	TBD Returns: `string`
`this.train ()` Full Usage: `this.train ()`	TBD
`this.backend` Full Usage: `this.backend` Returns: `Backend`	Returns: `Backend`
`this.state` Full Usage: `this.state` Returns: `ParameterDict`	TBD Returns: `ParameterDict`
`this.buffers` Full Usage: `this.buffers` Returns: `ParameterDict`	TBD Returns: `ParameterDict`
`this.hasOwnParameters` Full Usage: `this.hasOwnParameters` Returns: `bool`	TBD Returns: `bool`
`this.isNoDiff` Full Usage: `this.isNoDiff` Returns: `bool`	Returns: `bool`
`this.device` Full Usage: `this.device` Returns: `Device`	Returns: `Device`
`this.hasOwnBuffers` Full Usage: `this.hasOwnBuffers` Returns: `bool`	TBD Returns: `bool`
`this.parametersVector` Full Usage: `this.parametersVector` Returns: `Tensor`	TBD Returns: `Tensor`
`this.isReverseDiff` Full Usage: `this.isReverseDiff` Returns: `bool`	Returns: `bool`
`this.nstate` Full Usage: `this.nstate` Returns: `int`	TBD Returns: `int`
`this.nparameters` Full Usage: `this.nparameters` Returns: `int`	Gets the number of parameters of the Model Returns: `int`
`this.hasOwnState` Full Usage: `this.hasOwnState` Returns: `bool`	TBD Returns: `bool`
`this.children` Full Usage: `this.children` Returns: `ModelBase list`	TBD Returns: `ModelBase list`
`this.isForwardDiff` Full Usage: `this.isForwardDiff` Returns: `bool`	Returns: `bool`
`this.parameters` Full Usage: `this.parameters` Returns: `ParameterDict`	TBD Returns: `ParameterDict`
`this.descendants` Full Usage: `this.descendants` Returns: `ModelBase list`	TBD Returns: `ModelBase list`
`this.nbuffers` Full Usage: `this.nbuffers` Returns: `int`	TBD Returns: `int`
`this.buffersVector` Full Usage: `this.buffersVector` Returns: `Tensor`	TBD Returns: `Tensor`
`this.dtype` Full Usage: `this.dtype` Returns: `Dtype`	Returns: `Dtype`
`this.stateVector` Full Usage: `this.stateVector` Returns: `Tensor`	TBD Returns: `Tensor`

BatchNorm2d Type

Constructors

Instance members

Inherited members

Inherited from ModelBase

Instance members

Inherited from `ModelBase`