Package onnx

Class Onnx.TrainingInfoProto.Builder

  • All Implemented Interfaces:
    Cloneable, Onnx.TrainingInfoProtoOrBuilder, org.nd4j.shade.protobuf.Message.Builder, org.nd4j.shade.protobuf.MessageLite.Builder, org.nd4j.shade.protobuf.MessageLiteOrBuilder, org.nd4j.shade.protobuf.MessageOrBuilder
    Enclosing class:
    Onnx.TrainingInfoProto
    public static final class Onnx.TrainingInfoProto.Builder
extends org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>
implements Onnx.TrainingInfoProtoOrBuilder
     Training information
 TrainingInfoProto stores information for training a model.
 In particular, this defines two functionalities: an initialization-step
 and a training-algorithm-step. Initialization resets the model
 back to its original state as if no training has been performed.
 Training algorithm improves the model based on input data.
 The semantics of the initialization-step is that the initializers
 in ModelProto.graph and in TrainingInfoProto.algorithm are first
 initialized as specified by the initializers in the graph, and then
 updated by the "initialization_binding" in every instance in
 ModelProto.training_info.
 The field "algorithm" defines a computation graph which represents a
 training algorithm's step. After the execution of a
 TrainingInfoProto.algorithm, the initializers specified by "update_binding"
 may be immediately updated. If the targeted training algorithm contains
 consecutive update steps (such as block coordinate descent methods),
 the user needs to create a TrainingInfoProto for each step.
    Protobuf type onnx.TrainingInfoProto

    • Method Detail

      • getDescriptor

        public static final org.nd4j.shade.protobuf.Descriptors.Descriptor getDescriptor()

      • internalGetFieldAccessorTable

        protected org.nd4j.shade.protobuf.GeneratedMessageV3.FieldAccessorTable internalGetFieldAccessorTable()
        Specified by:
        internalGetFieldAccessorTable in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • clear

        public Onnx.TrainingInfoProto.Builder clear()
        Specified by:
        clear in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        clear in interface org.nd4j.shade.protobuf.MessageLite.Builder
        Overrides:
        clear in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • getDescriptorForType

        public org.nd4j.shade.protobuf.Descriptors.Descriptor getDescriptorForType()
        Specified by:
        getDescriptorForType in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        getDescriptorForType in interface org.nd4j.shade.protobuf.MessageOrBuilder
        Overrides:
        getDescriptorForType in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • getDefaultInstanceForType

        public Onnx.TrainingInfoProto getDefaultInstanceForType()
        Specified by:
        getDefaultInstanceForType in interface org.nd4j.shade.protobuf.MessageLiteOrBuilder
        Specified by:
        getDefaultInstanceForType in interface org.nd4j.shade.protobuf.MessageOrBuilder

      • build

        public Onnx.TrainingInfoProto build()
        Specified by:
        build in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        build in interface org.nd4j.shade.protobuf.MessageLite.Builder

      • buildPartial

        public Onnx.TrainingInfoProto buildPartial()
        Specified by:
        buildPartial in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        buildPartial in interface org.nd4j.shade.protobuf.MessageLite.Builder

      • clone

        public Onnx.TrainingInfoProto.Builder clone()
        Specified by:
        clone in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        clone in interface org.nd4j.shade.protobuf.MessageLite.Builder
        Overrides:
        clone in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • clearField

        public Onnx.TrainingInfoProto.Builder clearField​(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field)
        Specified by:
        clearField in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        clearField in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • clearOneof

        public Onnx.TrainingInfoProto.Builder clearOneof​(org.nd4j.shade.protobuf.Descriptors.OneofDescriptor oneof)
        Specified by:
        clearOneof in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        clearOneof in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • setRepeatedField

        public Onnx.TrainingInfoProto.Builder setRepeatedField​(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field,
                                                       int index,
                                                       Object value)
        Specified by:
        setRepeatedField in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        setRepeatedField in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • addRepeatedField

        public Onnx.TrainingInfoProto.Builder addRepeatedField​(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field,
                                                       Object value)
        Specified by:
        addRepeatedField in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        addRepeatedField in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • isInitialized

        public final boolean isInitialized()
        Specified by:
        isInitialized in interface org.nd4j.shade.protobuf.MessageLiteOrBuilder
        Overrides:
        isInitialized in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • mergeFrom

        public Onnx.TrainingInfoProto.Builder mergeFrom​(org.nd4j.shade.protobuf.CodedInputStream input,
                                                org.nd4j.shade.protobuf.ExtensionRegistryLite extensionRegistry)
                                         throws IOException
        Specified by:
        mergeFrom in interface org.nd4j.shade.protobuf.Message.Builder
        Specified by:
        mergeFrom in interface org.nd4j.shade.protobuf.MessageLite.Builder
        Overrides:
        mergeFrom in class org.nd4j.shade.protobuf.AbstractMessage.Builder<Onnx.TrainingInfoProto.Builder>
        Throws:
        IOException

      • hasInitialization

        public boolean hasInitialization()
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;
        Specified by:
        hasInitialization in interface Onnx.TrainingInfoProtoOrBuilder
        Returns:
        Whether the initialization field is set.

      • getInitialization

        public Onnx.GraphProto getInitialization()
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;
        Specified by:
        getInitialization in interface Onnx.TrainingInfoProtoOrBuilder
        Returns:
        The initialization.

      • setInitialization

        public Onnx.TrainingInfoProto.Builder setInitialization​(Onnx.GraphProto value)
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;

      • setInitialization

        public Onnx.TrainingInfoProto.Builder setInitialization​(Onnx.GraphProto.Builder builderForValue)
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;

      • mergeInitialization

        public Onnx.TrainingInfoProto.Builder mergeInitialization​(Onnx.GraphProto value)
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;

      • clearInitialization

        public Onnx.TrainingInfoProto.Builder clearInitialization()
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;

      • getInitializationBuilder

        public Onnx.GraphProto.Builder getInitializationBuilder()
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;

      • getInitializationOrBuilder

        public Onnx.GraphProtoOrBuilder getInitializationOrBuilder()
         This field describes a graph to compute the initial tensors
 upon starting the training process. Initialization graph has no input
 and can have multiple outputs. Usually, trainable tensors in neural
 networks are randomly initialized. To achieve that, for each tensor,
 the user can put a random number operator such as RandomNormal or
 RandomUniform in TrainingInfoProto.initialization.node and assign its
 random output to the specific tensor using "initialization_binding".
 This graph can also set the initializers in "algorithm" in the same
 TrainingInfoProto; a use case is resetting the number of training
 iteration to zero.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Thus, no initializer would be changed by default.
        .onnx.GraphProto initialization = 1;
        Specified by:
        getInitializationOrBuilder in interface Onnx.TrainingInfoProtoOrBuilder

      • hasAlgorithm

        public boolean hasAlgorithm()
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;
        Specified by:
        hasAlgorithm in interface Onnx.TrainingInfoProtoOrBuilder
        Returns:
        Whether the algorithm field is set.

      • getAlgorithm

        public Onnx.GraphProto getAlgorithm()
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;
        Specified by:
        getAlgorithm in interface Onnx.TrainingInfoProtoOrBuilder
        Returns:
        The algorithm.

      • setAlgorithm

        public Onnx.TrainingInfoProto.Builder setAlgorithm​(Onnx.GraphProto value)
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;

      • setAlgorithm

        public Onnx.TrainingInfoProto.Builder setAlgorithm​(Onnx.GraphProto.Builder builderForValue)
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;

      • mergeAlgorithm

        public Onnx.TrainingInfoProto.Builder mergeAlgorithm​(Onnx.GraphProto value)
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;

      • clearAlgorithm

        public Onnx.TrainingInfoProto.Builder clearAlgorithm()
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;

      • getAlgorithmBuilder

        public Onnx.GraphProto.Builder getAlgorithmBuilder()
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;

      • getAlgorithmOrBuilder

        public Onnx.GraphProtoOrBuilder getAlgorithmOrBuilder()
         This field represents a training algorithm step. Given required inputs,
 it computes outputs to update initializers in its own or inference graph's
 initializer lists. In general, this field contains loss node, gradient node,
 optimizer node, increment of iteration count.
 An execution of the training algorithm step is performed by executing the
 graph obtained by combining the inference graph (namely "ModelProto.graph")
 and the "algorithm" graph. That is, the actual the actual
 input/initializer/output/node/value_info/sparse_initializer list of
 the training graph is the concatenation of
 "ModelProto.graph.input/initializer/output/node/value_info/sparse_initializer"
 and "algorithm.input/initializer/output/node/value_info/sparse_initializer"
 in that order. This combined graph must satisfy the normal ONNX conditions.
 Now, let's provide a visualization of graph combination for clarity.
 Let the inference graph (i.e., "ModelProto.graph") be
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d
 and the "algorithm" graph be
    tensor_d -> Add -> tensor_e
 The combination process results
    tensor_a, tensor_b -> MatMul -> tensor_c -> Sigmoid -> tensor_d -> Add -> tensor_e
 Notice that an input of a node in the "algorithm" graph may reference the
 output of a node in the inference graph (but not the other way round). Also, inference
 node cannot reference inputs of "algorithm". With these restrictions, inference graph
 can always be run independently without training information.
 By default, this field is an empty graph and its evaluation does not
 produce any output. Evaluating the default training step never
 update any initializers.
        .onnx.GraphProto algorithm = 2;
        Specified by:
        getAlgorithmOrBuilder in interface Onnx.TrainingInfoProtoOrBuilder

      • getInitializationBindingList

        public List<Onnx.StringStringEntryProto> getInitializationBindingList()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;
        Specified by:
        getInitializationBindingList in interface Onnx.TrainingInfoProtoOrBuilder

      • getInitializationBindingCount

        public int getInitializationBindingCount()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;
        Specified by:
        getInitializationBindingCount in interface Onnx.TrainingInfoProtoOrBuilder

      • getInitializationBinding

        public Onnx.StringStringEntryProto getInitializationBinding​(int index)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;
        Specified by:
        getInitializationBinding in interface Onnx.TrainingInfoProtoOrBuilder

      • setInitializationBinding

        public Onnx.TrainingInfoProto.Builder setInitializationBinding​(int index,
                                                               Onnx.StringStringEntryProto value)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • setInitializationBinding

        public Onnx.TrainingInfoProto.Builder setInitializationBinding​(int index,
                                                               Onnx.StringStringEntryProto.Builder builderForValue)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addInitializationBinding

        public Onnx.TrainingInfoProto.Builder addInitializationBinding​(Onnx.StringStringEntryProto value)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addInitializationBinding

        public Onnx.TrainingInfoProto.Builder addInitializationBinding​(int index,
                                                               Onnx.StringStringEntryProto value)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addInitializationBinding

        public Onnx.TrainingInfoProto.Builder addInitializationBinding​(Onnx.StringStringEntryProto.Builder builderForValue)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addInitializationBinding

        public Onnx.TrainingInfoProto.Builder addInitializationBinding​(int index,
                                                               Onnx.StringStringEntryProto.Builder builderForValue)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addAllInitializationBinding

        public Onnx.TrainingInfoProto.Builder addAllInitializationBinding​(Iterable<? extends Onnx.StringStringEntryProto> values)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • clearInitializationBinding

        public Onnx.TrainingInfoProto.Builder clearInitializationBinding()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • removeInitializationBinding

        public Onnx.TrainingInfoProto.Builder removeInitializationBinding​(int index)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • getInitializationBindingBuilder

        public Onnx.StringStringEntryProto.Builder getInitializationBindingBuilder​(int index)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • getInitializationBindingOrBuilder

        public Onnx.StringStringEntryProtoOrBuilder getInitializationBindingOrBuilder​(int index)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;
        Specified by:
        getInitializationBindingOrBuilder in interface Onnx.TrainingInfoProtoOrBuilder

      • getInitializationBindingOrBuilderList

        public List<? extends Onnx.StringStringEntryProtoOrBuilder> getInitializationBindingOrBuilderList()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;
        Specified by:
        getInitializationBindingOrBuilderList in interface Onnx.TrainingInfoProtoOrBuilder

      • addInitializationBindingBuilder

        public Onnx.StringStringEntryProto.Builder addInitializationBindingBuilder()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • addInitializationBindingBuilder

        public Onnx.StringStringEntryProto.Builder addInitializationBindingBuilder​(int index)
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • getInitializationBindingBuilderList

        public List<Onnx.StringStringEntryProto.Builder> getInitializationBindingBuilderList()
         This field specifies the bindings from the outputs of "initialization" to
 some initializers in "ModelProto.graph.initializer" and
 the "algorithm.initializer" in the same TrainingInfoProto.
 See "update_binding" below for details.
 By default, this field is empty and no initializer would be changed
 by the execution of "initialization".
        repeated .onnx.StringStringEntryProto initialization_binding = 3;

      • getUpdateBindingList

        public List<Onnx.StringStringEntryProto> getUpdateBindingList()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;
        Specified by:
        getUpdateBindingList in interface Onnx.TrainingInfoProtoOrBuilder

      • getUpdateBindingCount

        public int getUpdateBindingCount()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;
        Specified by:
        getUpdateBindingCount in interface Onnx.TrainingInfoProtoOrBuilder

      • getUpdateBinding

        public Onnx.StringStringEntryProto getUpdateBinding​(int index)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;
        Specified by:
        getUpdateBinding in interface Onnx.TrainingInfoProtoOrBuilder

      • setUpdateBinding

        public Onnx.TrainingInfoProto.Builder setUpdateBinding​(int index,
                                                       Onnx.StringStringEntryProto value)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • setUpdateBinding

        public Onnx.TrainingInfoProto.Builder setUpdateBinding​(int index,
                                                       Onnx.StringStringEntryProto.Builder builderForValue)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addUpdateBinding

        public Onnx.TrainingInfoProto.Builder addUpdateBinding​(Onnx.StringStringEntryProto value)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addUpdateBinding

        public Onnx.TrainingInfoProto.Builder addUpdateBinding​(int index,
                                                       Onnx.StringStringEntryProto value)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addUpdateBinding

        public Onnx.TrainingInfoProto.Builder addUpdateBinding​(Onnx.StringStringEntryProto.Builder builderForValue)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addUpdateBinding

        public Onnx.TrainingInfoProto.Builder addUpdateBinding​(int index,
                                                       Onnx.StringStringEntryProto.Builder builderForValue)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addAllUpdateBinding

        public Onnx.TrainingInfoProto.Builder addAllUpdateBinding​(Iterable<? extends Onnx.StringStringEntryProto> values)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • clearUpdateBinding

        public Onnx.TrainingInfoProto.Builder clearUpdateBinding()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • removeUpdateBinding

        public Onnx.TrainingInfoProto.Builder removeUpdateBinding​(int index)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • getUpdateBindingBuilder

        public Onnx.StringStringEntryProto.Builder getUpdateBindingBuilder​(int index)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • getUpdateBindingOrBuilder

        public Onnx.StringStringEntryProtoOrBuilder getUpdateBindingOrBuilder​(int index)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;
        Specified by:
        getUpdateBindingOrBuilder in interface Onnx.TrainingInfoProtoOrBuilder

      • getUpdateBindingOrBuilderList

        public List<? extends Onnx.StringStringEntryProtoOrBuilder> getUpdateBindingOrBuilderList()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;
        Specified by:
        getUpdateBindingOrBuilderList in interface Onnx.TrainingInfoProtoOrBuilder

      • addUpdateBindingBuilder

        public Onnx.StringStringEntryProto.Builder addUpdateBindingBuilder()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • addUpdateBindingBuilder

        public Onnx.StringStringEntryProto.Builder addUpdateBindingBuilder​(int index)
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • getUpdateBindingBuilderList

        public List<Onnx.StringStringEntryProto.Builder> getUpdateBindingBuilderList()
         Gradient-based training is usually an iterative procedure. In one gradient
 descent iteration, we apply
 x = x - r * g
 where "x" is the optimized tensor, "r" stands for learning rate, and "g" is
 gradient of "x" with respect to a chosen loss. To avoid adding assignments
 into the training graph, we split the update equation into
 y = x - r * g
 x = y
 The user needs to save "y = x - r * g" into TrainingInfoProto.algorithm. To
 tell that "y" should be assigned to "x", the field "update_binding" may
 contain a key-value pair of strings, "x" (key of StringStringEntryProto)
 and "y" (value of StringStringEntryProto).
 For a neural network with multiple trainable (mutable) tensors, there can
 be multiple key-value pairs in "update_binding".
 The initializers appears as keys in "update_binding" are considered
 mutable variables. This implies some behaviors
 as described below.
  1. We have only unique keys in all "update_binding"s so that two
     variables may not have the same name. This ensures that one
     variable is assigned up to once.
  2. The keys must appear in names of "ModelProto.graph.initializer" or
     "TrainingInfoProto.algorithm.initializer".
  3. The values must be output names of "algorithm" or "ModelProto.graph.output".
  4. Mutable variables are initialized to the value specified by the
     corresponding initializer, and then potentially updated by
     "initializer_binding"s and "update_binding"s in "TrainingInfoProto"s.
 This field usually contains names of trainable tensors
 (in ModelProto.graph), optimizer states such as momentums in advanced
 stochastic gradient methods (in TrainingInfoProto.graph),
 and number of training iterations (in TrainingInfoProto.graph).
 By default, this field is empty and no initializer would be changed
 by the execution of "algorithm".
        repeated .onnx.StringStringEntryProto update_binding = 4;

      • setUnknownFields

        public final Onnx.TrainingInfoProto.Builder setUnknownFields​(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
        Specified by:
        setUnknownFields in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        setUnknownFields in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>

      • mergeUnknownFields

        public final Onnx.TrainingInfoProto.Builder mergeUnknownFields​(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
        Specified by:
        mergeUnknownFields in interface org.nd4j.shade.protobuf.Message.Builder
        Overrides:
        mergeUnknownFields in class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<Onnx.TrainingInfoProto.Builder>