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

MachineLearningCommandJob 类

定义

命令作业定义。

public class MachineLearningCommandJob : Azure.ResourceManager.MachineLearning.Models.MachineLearningJobProperties
type MachineLearningCommandJob = class
    inherit MachineLearningJobProperties
Public Class MachineLearningCommandJob
Inherits MachineLearningJobProperties
继承

构造函数

MachineLearningCommandJob(String, ResourceIdentifier)

初始化 MachineLearningCommandJob 的新实例。

属性

CodeId

代码资产的 ARM 资源 ID。

Command

[必需]作业启动时要执行的命令。 例如 “python train.py”。

ComponentId

组件资源的 ARM 资源 ID。

(继承自 MachineLearningJobProperties)
ComputeId

计算资源的 ARM 资源 ID。

(继承自 MachineLearningJobProperties)
Description

资产说明文本。

(继承自 MachineLearningResourceBase)
DisplayName

作业的显示名称。

(继承自 MachineLearningJobProperties)
Distribution

作业的分发配置。 如果已设置,则应为 Mpi、Tensorflow、PyTorch、Ray 或 null 之一。 请注意 MachineLearningDistributionConfiguration , 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 MpiDistributionConfigurationPyTorchDistributionConfigurationRayDistributionConfigurationTensorFlowDistributionConfiguration

EnvironmentId

[必需]作业的环境规范的 ARM 资源 ID。

EnvironmentVariables

作业中包含的环境变量。

ExperimentName

作业所属的试验的名称。 如果未设置,作业将置于“默认”试验中。

(继承自 MachineLearningJobProperties)
Identity

标识配置。 如果已设置,应为 AmlToken、ManagedIdentity、UserIdentity 或 null 之一。 如果为 null,则默认为 AmlToken。 请注意 MachineLearningIdentityConfiguration , 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 AmlTokenMachineLearningManagedIdentityMachineLearningUserIdentity

(继承自 MachineLearningJobProperties)
Inputs

作业中使用的输入数据绑定的映射。 请注意 MachineLearningJobInput , 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 MachineLearningCustomModelJobInput、、MachineLearningLiteralJobInputMachineLearningFlowModelJobInputMachineLearningTableJobInputMachineLearningTritonModelJobInputMachineLearningUriFileJobInputMachineLearningUriFolderJobInput

IsArchived

资产是否已存档?。

(继承自 MachineLearningJobProperties)
Limits

命令作业限制。

MlflowAutologger

[必需]指示是否启用 mlflow 自动记录器。

NotificationSetting

作业的通知设置。

(继承自 MachineLearningJobProperties)
Outputs

作业中使用的输出数据绑定的映射。 请注意 MachineLearningJobOutput , 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 MachineLearningCustomModelJobOutput、、MachineLearningFlowModelJobOutputMachineLearningTableJobOutputMachineLearningTritonModelJobOutputMachineLearningUriFileJobOutputMachineLearningUriFolderJobOutput

Parameters

输入参数。

若要将对象分配给此属性,请使用 FromObjectAsJson<T>(T, JsonSerializerOptions)

若要将已格式化的 json 字符串分配给此属性,请使用 FromString(String)

示例:

  • BinaryData.FromObjectAsJson (“foo”) 创建“foo”的有效负载。
  • BinaryData.FromString (“\”foo\“”) 创建“foo”的有效负载。
  • BinaryData.FromObjectAsJson (new { key = “value” }) 创建 { “key”: “value” } 的有效负载。
  • BinaryData.FromString (“{\”key\“: \”value\“}”) 创建 { “key”: “value” } 的有效负载。

Properties

资产属性字典。

(继承自 MachineLearningResourceBase)
QueueSettings

作业的队列设置。

Resources

作业的计算资源配置。

SecretsConfiguration

配置在运行时可用的机密。

(继承自 MachineLearningJobProperties)
Services

JobEndpoint 列表。 对于本地作业,作业终结点的终结点值为 FileStreamObject。

(继承自 MachineLearningJobProperties)
Status

作业的状态。

(继承自 MachineLearningJobProperties)
Tags

标记字典。 可以添加、删除和更新标记。

(继承自 MachineLearningResourceBase)

适用于