create_dl_layer_loss_huberT_create_dl_layer_loss_huberCreateDlLayerLossHuberCreateDlLayerLossHubercreate_dl_layer_loss_huber (Operator)

Name

create_dl_layer_loss_huberT_create_dl_layer_loss_huberCreateDlLayerLossHuberCreateDlLayerLossHubercreate_dl_layer_loss_huber — Create a Huber loss layer.

Signature

Description

The operator create_dl_layer_loss_hubercreate_dl_layer_loss_huberCreateDlLayerLossHuberCreateDlLayerLossHuberCreateDlLayerLossHubercreate_dl_layer_loss_huber creates a Huber loss layer whose handle is returned in DLLayerLossHuberDLLayerLossHuberDLLayerLossHuberDLLayerLossHuberDLLayerLossHuberdllayer_loss_huber. The Huber loss is defined by

This layer expects multiple layers as input:

• DLLayerInputDLLayerInputDLLayerInputDLLayerInputDLLayerInputdllayer_input: Specifies (usually a softmax layer).

• DLLayerTargetDLLayerTargetDLLayerTargetDLLayerTargetDLLayerTargetdllayer_target: Specifies the targets .

• DLLayerWeightsDLLayerWeightsDLLayerWeightsDLLayerWeightsDLLayerWeightsdllayer_weights: Specifies the weights .

The underlying data tensors are assumed to be of the same shape with a total number of elements.

The parameter DLLayerNormalizationDLLayerNormalizationDLLayerNormalizationDLLayerNormalizationDLLayerNormalizationdllayer_normalization can be used to determine the normalization factor . If DLLayerNormalizationDLLayerNormalizationDLLayerNormalizationDLLayerNormalizationDLLayerNormalizationdllayer_normalization is set to an empty tuple, the sum over all weights is used for the normalization .

The parameter LossWeightLossWeightLossWeightLossWeightlossWeightloss_weight determines the scalar weight factor .

The parameter BetaBetaBetaBetabetabeta sets the value for in the formula. If BetaBetaBetaBetabetabeta is set to 0, the Huber loss is equal to an L1-loss.

The parameter LayerNameLayerNameLayerNameLayerNamelayerNamelayer_name sets an individual layer name. Note that if creating a model using create_dl_modelcreate_dl_modelCreateDlModelCreateDlModelCreateDlModelcreate_dl_model each layer of the created network must have a unique name.

The following generic parameters GenParamNameGenParamNameGenParamNameGenParamNamegenParamNamegen_param_name and the corresponding values GenParamValueGenParamValueGenParamValueGenParamValuegenParamValuegen_param_value are supported:

'is_inference_output'"is_inference_output""is_inference_output""is_inference_output""is_inference_output""is_inference_output":

Determines whether apply_dl_modelapply_dl_modelApplyDlModelApplyDlModelApplyDlModelapply_dl_model will include the output of this layer in the dictionary DLResultBatchDLResultBatchDLResultBatchDLResultBatchDLResultBatchdlresult_batch even without specifying this layer in OutputsOutputsOutputsOutputsoutputsoutputs ('true'"true""true""true""true""true") or not ('false'"false""false""false""false""false").

Default: 'false'"false""false""false""false""false"

Certain parameters of layers created using this operator create_dl_layer_loss_hubercreate_dl_layer_loss_huberCreateDlLayerLossHuberCreateDlLayerLossHuberCreateDlLayerLossHubercreate_dl_layer_loss_huber can be set and retrieved using further operators. The following tables give an overview, which parameters can be set using set_dl_model_layer_paramset_dl_model_layer_paramSetDlModelLayerParamSetDlModelLayerParamSetDlModelLayerParamset_dl_model_layer_param and which ones can be retrieved using get_dl_model_layer_paramget_dl_model_layer_paramGetDlModelLayerParamGetDlModelLayerParamGetDlModelLayerParamget_dl_model_layer_param or get_dl_layer_paramget_dl_layer_paramGetDlLayerParamGetDlLayerParamGetDlLayerParamget_dl_layer_param. Note, the operators set_dl_model_layer_paramset_dl_model_layer_paramSetDlModelLayerParamSetDlModelLayerParamSetDlModelLayerParamset_dl_model_layer_param and get_dl_model_layer_paramget_dl_model_layer_paramGetDlModelLayerParamGetDlModelLayerParamGetDlModelLayerParamget_dl_model_layer_param require a model created by create_dl_modelcreate_dl_modelCreateDlModelCreateDlModelCreateDlModelcreate_dl_model.

Execution Information

• Multithreading type: reentrant (runs in parallel with non-exclusive operators).
• Multithreading scope: global (may be called from any thread).
• Processed without parallelization.

Parameters

Module

Deep Learning Training