通过

将计算管理升级到 v2

  • 项目

v2 开发平台的计算管理功能保持不变。

本文比较了 SDK v1 和 SDK v2 中的方案。

创建计算实例

  • SDK v1

    import datetime
import time

from azureml.core.compute import ComputeTarget, ComputeInstance
from azureml.core.compute_target import ComputeTargetException

# Compute Instances need to have a unique name across the region.
# Here, we create a unique name with current datetime
ci_basic_name = "basic-ci" + datetime.datetime.now().strftime("%Y%m%d%H%M")

compute_config = ComputeInstance.provisioning_configuration(
        vm_size='STANDARD_DS3_V2'
    )
    instance = ComputeInstance.create(ws, ci_basic_name , compute_config)
    instance.wait_for_completion(show_output=True)

  • SDK v2

    # Compute Instances need to have a unique name across the region.
# Here, we create a unique name with current datetime
from azure.ai.ml.entities import ComputeInstance, AmlCompute
import datetime

ci_basic_name = "basic-ci" + datetime.datetime.now().strftime("%Y%m%d%H%M")
ci_basic = ComputeInstance(name=ci_basic_name, size="STANDARD_DS3_v2", idle_time_before_shutdown_minutes="30")
ml_client.begin_create_or_update(ci_basic)

创建计算群集

  • SDK v1

    from azureml.core.compute import ComputeTarget, AmlCompute
from azureml.core.compute_target import ComputeTargetException

# Choose a name for your CPU cluster
cpu_cluster_name = "cpucluster"
compute_config = AmlCompute.provisioning_configuration(vm_size='STANDARD_DS3_V2',
                                                           max_nodes=4)
cpu_cluster = ComputeTarget.create(ws, cpu_cluster_name, compute_config)
cpu_cluster.wait_for_completion(show_output=True)

  • SDK v2

    from azure.ai.ml.entities import AmlCompute
cpu_cluster_name = "cpucluster"
cluster_basic = AmlCompute(
    name=cpu_cluster_name,
    type="amlcompute",
    size="STANDARD_DS3_v2",
    max_instances=4
)
ml_client.begin_create_or_update(cluster_basic)

SDK v1 和 SDK v2 中关键功能的映射

SDK v1 中的功能 SDK v2 中的粗略映射
SDK v1 中的方法/API（使用指向参考文档的链接） SDK v2 中的方法/API（使用指向参考文档的链接）

后续步骤