你当前正在访问 Microsoft Azure Global Edition 技术文档网站。如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站，请访问 https://docs.azure.cn。

ArmMachineLearningModelFactory.MachineLearningBatchDeploymentProperties 方法

参考

定义

命名空间:: Azure.ResourceManager.MachineLearning.Models

程序集:: Azure.ResourceManager.MachineLearning.dll

包:: Azure.ResourceManager.MachineLearning v1.2.0-beta.2

重要

一些信息与预发行产品相关，相应产品在发行之前可能会进行重大修改。对于此处提供的信息，Microsoft 不作任何明示或暗示的担保。

初始化 MachineLearningBatchDeploymentProperties 的新实例。

public static Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchDeploymentProperties MachineLearningBatchDeploymentProperties (Azure.ResourceManager.MachineLearning.Models.MachineLearningCodeConfiguration codeConfiguration = default, string description = default, string environmentId = default, System.Collections.Generic.IDictionary<string,string> environmentVariables = default, System.Collections.Generic.IDictionary<string,string> properties = default, string compute = default, Azure.ResourceManager.MachineLearning.Models.BatchDeploymentConfiguration deploymentConfiguration = default, int? errorThreshold = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchLoggingLevel? loggingLevel = default, int? maxConcurrencyPerInstance = default, long? miniBatchSize = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningAssetReferenceBase model = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchOutputAction? outputAction = default, string outputFileName = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningDeploymentProvisioningState? provisioningState = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningDeploymentResourceConfiguration resources = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchRetrySettings retrySettings = default);

static member MachineLearningBatchDeploymentProperties : Azure.ResourceManager.MachineLearning.Models.MachineLearningCodeConfiguration * string * string * System.Collections.Generic.IDictionary<string, string> * System.Collections.Generic.IDictionary<string, string> * string * Azure.ResourceManager.MachineLearning.Models.BatchDeploymentConfiguration * Nullable<int> * Nullable<Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchLoggingLevel> * Nullable<int> * Nullable<int64> * Azure.ResourceManager.MachineLearning.Models.MachineLearningAssetReferenceBase * Nullable<Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchOutputAction> * string * Nullable<Azure.ResourceManager.MachineLearning.Models.MachineLearningDeploymentProvisioningState> * Azure.ResourceManager.MachineLearning.Models.MachineLearningDeploymentResourceConfiguration * Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchRetrySettings -> Azure.ResourceManager.MachineLearning.Models.MachineLearningBatchDeploymentProperties

Public Shared Function MachineLearningBatchDeploymentProperties (Optional codeConfiguration As MachineLearningCodeConfiguration = Nothing, Optional description As String = Nothing, Optional environmentId As String = Nothing, Optional environmentVariables As IDictionary(Of String, String) = Nothing, Optional properties As IDictionary(Of String, String) = Nothing, Optional compute As String = Nothing, Optional deploymentConfiguration As BatchDeploymentConfiguration = Nothing, Optional errorThreshold As Nullable(Of Integer) = Nothing, Optional loggingLevel As Nullable(Of MachineLearningBatchLoggingLevel) = Nothing, Optional maxConcurrencyPerInstance As Nullable(Of Integer) = Nothing, Optional miniBatchSize As Nullable(Of Long) = Nothing, Optional model As MachineLearningAssetReferenceBase = Nothing, Optional outputAction As Nullable(Of MachineLearningBatchOutputAction) = Nothing, Optional outputFileName As String = Nothing, Optional provisioningState As Nullable(Of MachineLearningDeploymentProvisioningState) = Nothing, Optional resources As MachineLearningDeploymentResourceConfiguration = Nothing, Optional retrySettings As MachineLearningBatchRetrySettings = Nothing) As MachineLearningBatchDeploymentProperties

参数

codeConfiguration: MachineLearningCodeConfiguration

终结点部署的代码配置。

description: String

终结点部署的说明。

environmentId: String

终结点部署的环境规范的 ARM 资源 ID。

environmentVariables: IDictionary<String,String>

部署的环境变量配置。

properties: IDictionary<String,String>

属性字典。可以添加属性，但不能删除或更改。

compute: String

批量推理操作的计算目标。

deploymentConfiguration: BatchDeploymentConfiguration

与不同部署类型相关的属性。请注意 BatchDeploymentConfiguration ，是基类。根据方案，可能需要在此处分配基类的派生类，或者需要将此属性强制转换为可能的派生类之一。可用的派生类包括 BatchPipelineComponentDeploymentConfiguration。

errorThreshold: Nullable<Int32>

错误阈值，如果整个输入的错误计数超过此值，则批处理推理将中止。范围为 [-1， int.MaxValue]。对于 FileDataset，此值是文件失败的计数。对于 TabularDataset，此值是记录失败的计数。如果设置为 -1 (下限) ，则将忽略批处理推理期间的所有失败。

loggingLevel: Nullable<MachineLearningBatchLoggingLevel>

批量推理操作的日志记录级别。

maxConcurrencyPerInstance: Nullable<Int32>

指示每个实例的最大并行数。

miniBatchSize: Nullable<Int64>

传递给每个批处理调用的微型批的大小。对于 FileDataset，这是每个微型批的文件数。对于 TabularDataset，这是每个微批记录的大小（以字节为单位）。

model: MachineLearningAssetReferenceBase

对终结点部署的模型资产的引用。请注意 MachineLearningAssetReferenceBase ，是基类。根据方案，可能需要在此处分配基类的派生类，或者需要将此属性强制转换为可能的派生类之一。可用的派生类包括 MachineLearningDataPathAssetReference、 MachineLearningOutputPathAssetReference 和 MachineLearningIdAssetReference。