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.MachineLearningSweepJob 方法
定义
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 MachineLearningSweepJob 的新实例。
public static Azure.ResourceManager.MachineLearning.Models.MachineLearningSweepJob MachineLearningSweepJob (string description = default, System.Collections.Generic.IDictionary<string,string> properties = default, System.Collections.Generic.IDictionary<string,string> tags = default, Azure.Core.ResourceIdentifier componentId = default, Azure.Core.ResourceIdentifier computeId = default, string displayName = default, string experimentName = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningIdentityConfiguration identity = default, bool? isArchived = default, Azure.ResourceManager.MachineLearning.Models.NotificationSetting notificationSetting = default, System.Collections.Generic.IDictionary<string,Azure.ResourceManager.MachineLearning.Models.SecretConfiguration> secretsConfiguration = default, System.Collections.Generic.IDictionary<string,Azure.ResourceManager.MachineLearning.Models.MachineLearningJobService> services = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningJobStatus? status = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningEarlyTerminationPolicy earlyTermination = default, System.Collections.Generic.IDictionary<string,Azure.ResourceManager.MachineLearning.Models.MachineLearningJobInput> inputs = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningSweepJobLimits limits = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningObjective objective = default, System.Collections.Generic.IDictionary<string,Azure.ResourceManager.MachineLearning.Models.MachineLearningJobOutput> outputs = default, Azure.ResourceManager.MachineLearning.Models.JobQueueSettings queueSettings = default, Azure.ResourceManager.MachineLearning.Models.SamplingAlgorithm samplingAlgorithm = default, BinaryData searchSpace = default, Azure.ResourceManager.MachineLearning.Models.MachineLearningTrialComponent trial = default);
static member MachineLearningSweepJob : string * System.Collections.Generic.IDictionary<string, string> * System.Collections.Generic.IDictionary<string, string> * Azure.Core.ResourceIdentifier * Azure.Core.ResourceIdentifier * string * string * Azure.ResourceManager.MachineLearning.Models.MachineLearningIdentityConfiguration * Nullable<bool> * Azure.ResourceManager.MachineLearning.Models.NotificationSetting * System.Collections.Generic.IDictionary<string, Azure.ResourceManager.MachineLearning.Models.SecretConfiguration> * System.Collections.Generic.IDictionary<string, Azure.ResourceManager.MachineLearning.Models.MachineLearningJobService> * Nullable<Azure.ResourceManager.MachineLearning.Models.MachineLearningJobStatus> * Azure.ResourceManager.MachineLearning.Models.MachineLearningEarlyTerminationPolicy * System.Collections.Generic.IDictionary<string, Azure.ResourceManager.MachineLearning.Models.MachineLearningJobInput> * Azure.ResourceManager.MachineLearning.Models.MachineLearningSweepJobLimits * Azure.ResourceManager.MachineLearning.Models.MachineLearningObjective * System.Collections.Generic.IDictionary<string, Azure.ResourceManager.MachineLearning.Models.MachineLearningJobOutput> * Azure.ResourceManager.MachineLearning.Models.JobQueueSettings * Azure.ResourceManager.MachineLearning.Models.SamplingAlgorithm * BinaryData * Azure.ResourceManager.MachineLearning.Models.MachineLearningTrialComponent -> Azure.ResourceManager.MachineLearning.Models.MachineLearningSweepJob
Public Shared Function MachineLearningSweepJob (Optional description As String = Nothing, Optional properties As IDictionary(Of String, String) = Nothing, Optional tags As IDictionary(Of String, String) = Nothing, Optional componentId As ResourceIdentifier = Nothing, Optional computeId As ResourceIdentifier = Nothing, Optional displayName As String = Nothing, Optional experimentName As String = Nothing, Optional identity As MachineLearningIdentityConfiguration = Nothing, Optional isArchived As Nullable(Of Boolean) = Nothing, Optional notificationSetting As NotificationSetting = Nothing, Optional secretsConfiguration As IDictionary(Of String, SecretConfiguration) = Nothing, Optional services As IDictionary(Of String, MachineLearningJobService) = Nothing, Optional status As Nullable(Of MachineLearningJobStatus) = Nothing, Optional earlyTermination As MachineLearningEarlyTerminationPolicy = Nothing, Optional inputs As IDictionary(Of String, MachineLearningJobInput) = Nothing, Optional limits As MachineLearningSweepJobLimits = Nothing, Optional objective As MachineLearningObjective = Nothing, Optional outputs As IDictionary(Of String, MachineLearningJobOutput) = Nothing, Optional queueSettings As JobQueueSettings = Nothing, Optional samplingAlgorithm As SamplingAlgorithm = Nothing, Optional searchSpace As BinaryData = Nothing, Optional trial As MachineLearningTrialComponent = Nothing) As MachineLearningSweepJob
参数
- description
- String
资产说明文本。
- properties
- IDictionary<String,String>
资产属性字典。
- tags
- IDictionary<String,String>
标记字典。 可以添加、删除和更新标记。
- componentId
- ResourceIdentifier
组件资源的 ARM 资源 ID。
- computeId
- ResourceIdentifier
计算资源的 ARM 资源 ID。
- displayName
- String
作业的显示名称。
- experimentName
- String
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。
- identity
- MachineLearningIdentityConfiguration
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 AmlToken、 MachineLearningManagedIdentity 和 MachineLearningUserIdentity。
- notificationSetting
- NotificationSetting
作业的通知设置。
- secretsConfiguration
- IDictionary<String,SecretConfiguration>
配置在运行时可用的机密。
- services
- IDictionary<String,MachineLearningJobService>
JobEndpoint 列表。 对于本地作业,作业终结点的终结点值为 FileStreamObject。
- status
- Nullable<MachineLearningJobStatus>
作业的状态。
- earlyTermination
- MachineLearningEarlyTerminationPolicy
提前终止策略允许在完成之前取消性能不佳的运行。请注意 MachineLearningEarlyTerminationPolicy 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 BanditPolicy、 MedianStoppingPolicy 和 TruncationSelectionPolicy。
- inputs
- IDictionary<String,MachineLearningJobInput>
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、、MachineLearningLiteralJobInput、MachineLearningFlowModelJobInputMachineLearningTableJobInput、 MachineLearningTritonModelJobInputMachineLearningUriFileJobInput 和 MachineLearningUriFolderJobInput。
扫描作业限制。
- objective
- MachineLearningObjective
[必需]优化目标。
- outputs
- IDictionary<String,MachineLearningJobOutput>
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、、MachineLearningFlowModelJobOutputMachineLearningTableJobOutput、 MachineLearningTritonModelJobOutputMachineLearningUriFileJobOutput 和 MachineLearningUriFolderJobOutput。
- queueSettings
- JobQueueSettings
作业的队列设置。
- samplingAlgorithm
- SamplingAlgorithm
[必需]超参数采样算法请注意 SamplingAlgorithm 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 BayesianSamplingAlgorithm、 GridSamplingAlgorithm 和 RandomSamplingAlgorithm。
- searchSpace
- BinaryData
[必需]包含每个参数及其分布的字典。 字典键是参数的名称。
[必需]试用组件定义。
返回
用于模拟的新 MachineLearningSweepJob 实例。