Methods#
Summary#
method 
class parent 
truncated documentation 

BaseEstimator 
Removes any non pickable attribute. 

BaseLearningOnnx 
Overwrites getstate to get rid of InferenceSession. 

OrtDataLoader 
Removes any non pickable attribute. 

OrtGradientForwardBackwardOptimizer 
Removes any non pickable attribute. 

OrtGradientForwardBackward 
Removes any non pickable attribute. 

BaseEstimator 

BaseLearningOnnx 

OrtDataLoader 

OrtGradientOptimizer 

OrtGradientForwardBackwardOptimizer 

OrtGradientForwardBackward 

OrtGradientForwardBackwardFunction 

AbsoluteLearningLoss 

BaseLearningLoss 

ElasticLearningLoss 

NegLogLearningLoss 

SquareLearningLoss 

BaseLearningPenalty 

ElasticLearningPenalty 

NoLearningPenalty 

BaseLearningRate 

LearningRateSGD 

LearningRateSGDNesterov 

OnnxSegment 

OnnxSplitting 

OrtDataLoader 
Returns the number of observations. 

BaseEstimator 
Usual. 

BaseLearningOnnx 
Usual 

OrtDataLoader 
usual 

OrtGradientForwardBackward 
usual 

OnnxSegment 

BaseLearningOnnx 

BaseLearningRate 


LearningRateSGD 


LearningRateSGDNesterov 

BaseEstimator 
Restores any non pickable attribute. 

BaseLearningOnnx 
Overwrites getstate to get rid of InferenceSession. 

OrtDataLoader 
Restores any non pickable attribute. 

OrtGradientForwardBackwardOptimizer 
Restores any non pickable attribute. 

OrtGradientForwardBackward 
Restores any non pickable attribute. 

BaseLearningOnnx 
Binds C_OrtValue to the structure used by InferenceSession to run inference. 

OrtGradientOptimizer 
Binds C_OrtValue to the structure used by InferenceSession to run inference. 

BaseLearningOnnx 
Binds C_OrtValue to the structure used by InferenceSession to run inference. 


AbsoluteLearningLoss 

BaseLearningLoss 


ElasticLearningLoss 


NegLogLearningLoss 


SquareLearningLoss 

BaseLearningPenalty 


ElasticLearningPenalty 


NoLearningPenalty 

BaseLearningRate 


LearningRateSGD 


LearningRateSGDNesterov 

OrtGradientForwardBackward 
Creates forward and backward ONNX graph. The new class has the following attributes: 

OrtGradientOptimizer 
Creates an instance of TrainingSession. 

OrtGradientForwardBackwardOptimizer 

OrtGradientOptimizer 

OrtGradientForwardBackwardOptimizer 

OrtGradientForwardBackwardOptimizer 

OnnxSplitting 

BaseEstimator 
Returns the trained onnx graph, the initial graph modified by replacing the initializers with the trained … 

OnnxSplitting 

OrtGradientForwardBackward 

OrtGradientOptimizer 

OrtGradientForwardBackwardOptimizer 

OnnxSplitting 
Builds one onnx subpart including segments from a to b (excluded). 

OnnxSplitting 

OrtDataLoader 

OnnxSplitting 
Splits the segments into two groups of the same size. 

OrtGradientForwardBackwardFunction 
Implements backward function. The function returns an OrtValueVector. 

BaseLearningOnnx 
This class computes a function represented as an ONNX graph. This method builds it. This function creates … 

OrtGradientForwardBackwardOptimizer 
Creates ONNX graph and InferenceSession related to any operations applying on OrtValue. 

AbsoluteLearningLoss 

ElasticLearningLoss 

NegLogLearningLoss 

SquareLearningLoss 

ElasticLearningPenalty 

NoLearningPenalty 

LearningRateSGD 

LearningRateSGDNesterov 


AbsoluteLearningLoss 
Assuming the loss function was created. This one takes the onnx graph and generate the onnx graph for the … 
BaseLearningLoss 
Assuming the loss function was created. This one takes the onnx graph and generate the onnx graph for the … 


ElasticLearningLoss 
Assuming the loss function was created. This one takes the onnx graph and generate the onnx graph for the … 

NegLogLearningLoss 
Assuming the loss function was created. This one takes the onnx graph and generate the onnx graph for the … 

SquareLearningLoss 
Assuming the loss function was created. This one takes the onnx graph and generate the onnx graph for the … 
BaseLearningOnnx 
Clears binding and empty cache. 

OrtGradientOptimizer 
Trains the model. 

OrtGradientForwardBackwardOptimizer 
Trains the model. 

OrtGradientForwardBackwardFunction 
Implements forward function. 

OrtGradientForwardBackwardOptimizer 
Returns the trained weights and the inputs. 

OrtGradientForwardBackward 
Returns an initializer as numpy arrays. 

BaseEstimator 
Returns the list of parameters. Parameter deep is unused. 

OrtGradientOptimizer 
Returns the trained weights. 

OrtGradientForwardBackwardOptimizer 
Returns the trained weights. 

BaseEstimator 
Returns the trained onnx graph, the initial graph modified by replacing the initializers with the trained … 

OrtGradientOptimizer 
Returns the trained onnx graph, the initial graph modified by replacing the initializers with the trained … 

OrtGradientForwardBackwardOptimizer 
Returns the trained onnx graph, the initial graph modified by replacing the initializers with the trained … 

BaseLearningRate 
Initializes the learning rate at the beginning of the training. 

LearningRateSGD 
Updates the learning rate at the end of an iteration. 

LearningRateSGDNesterov 
Updates the learning rate at the end of an iteration. 

OrtDataLoader 
Iterates over the datasets by drawing batch_size consecutive observations. Modifies a bind structure. 

OrtDataLoader 
Iterates over the datasets by drawing batch_size consecutive observations. This iterator is slow as it … 

OrtDataLoader 
Iterates over the datasets by drawing batch_size consecutive observations. This iterator is slow as it … 

BaseLearningRate 
Loops over learning rate values, n to be precise. 


LearningRateSGD 
Loops over learning rate values, n to be precise. 

LearningRateSGDNesterov 
Loops over learning rate values, n to be precise. 

AbsoluteLearningLoss 
Returns the loss and the gradient as OrtValue. 
BaseLearningLoss 
Returns the loss and the gradient as OrtValue. 


ElasticLearningLoss 
Returns the loss and the gradient as OrtValue. 

NegLogLearningLoss 
Returns the loss and the gradient as OrtValue. 

SquareLearningLoss 
Returns the loss and the gradient as OrtValue. 

AbsoluteLearningLoss 
Returns the weighted loss (or score) for every observation as OrtValue. 
BaseLearningLoss 
Returns the weighted loss (or score) for every observation as OrtValue. 


ElasticLearningLoss 
Returns the weighted loss (or score) for every observation as OrtValue. 

NegLogLearningLoss 
Returns the weighted loss (or score) for every observation as OrtValue. 

SquareLearningLoss 
Returns the weighted loss (or score) for every observation as OrtValue. 
OrtGradientForwardBackwardOptimizer 
Returns the losses associated to every observation. 

OnnxSplitting 
Builds onnx subparts based on the segmentation defined by extremities. 

OrtGradientForwardBackward 
Creates an instance of class self.cls_type_. It implements methods forward and backward. 

BaseLearningPenalty 
Returns the received loss. Updates the loss inplace. 

ElasticLearningPenalty 
Computes the penalty associated to every weights and adds them up to the loss. 

NoLearningPenalty 
Returns the received loss. Updates the loss inplace. 

OrtGradientForwardBackwardFunction 
Saves inputs furing forward steps. The list inputs is copied (simple copy, no deep copy). 

BaseOnnxClass 
Saves all ONNX files stored in this class. 

OrtGradientForwardBackwardOptimizer 
Return the whole score associated. 

BaseEstimator 
Returns the list of parameters. Parameter deep is unused. 

OrtGradientOptimizer 
Changes the trained weights. 

OrtGradientForwardBackwardOptimizer 
Changes the trained weights. 

OnnxSplitting 
Splits the segments into n_parts segments 

BaseLearningRate 
Updates the learning rate at the end of an iteration. 

LearningRateSGD 
Updates the learning rate at the end of an iteration. 

LearningRateSGDNesterov 
Updates the learning rate at the end of an iteration. 

BaseLearningPenalty 
Returns the received loss. Updates the weight inplace. 

ElasticLearningPenalty 

NoLearningPenalty 
Returns the received loss. Updates the weight inplace. 

BaseLearningRate 
Updates weights based on the algorithm this class is setting up. 

LearningRateSGD 

LearningRateSGDNesterov 