ParentT
- the stage of the parent definition to return to after attaching this definitionpublic static interface ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> extends Attachable.InDefinition<ParentT>, ToolTypeSettings.DefinitionStages.WithPythonInterpreter<ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT>>
Modifier and Type | Method and Description |
---|---|
ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> |
withParameterServerCommandLineArgs(String commandLineArgs) |
ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> |
withParameterServerCount(int parameterServerCount) |
ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> |
withWorkerCommandLineArgs(String commandLineArgs) |
ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> |
withWorkerCount(int workerCount) |
attach
withPythonInterpreterPath
ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> withWorkerCommandLineArgs(String commandLineArgs)
commandLineArgs
- specifies the command line arguments for the worker task.
This property is optional for single machine training.ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> withParameterServerCommandLineArgs(String commandLineArgs)
commandLineArgs
- specifies the command line arguments for the parameter server task.
This property is optional for single machine training.ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> withWorkerCount(int workerCount)
workerCount
- the number of worker tasks.
If specified, the value must be less than or equal to (nodeCount *
numberOfGPUs per VM). If not specified, the default value is equal to
nodeCount. This property can be specified only for distributed
TensorFlow training.ToolTypeSettings.TensorFlow.DefinitionStages.WithAttach<ParentT> withParameterServerCount(int parameterServerCount)
parameterServerCount
- the number of parameter server tasks.
If specified, the value must be less than or equal to nodeCount. If not
specified, the default value is equal to 1 for distributed TensorFlow
training (This property is not applicable for single machine training).
This property can be specified only for distributed TensorFlow training.Copyright © 2019. All rights reserved.