Azure ML ParallelRunStep导入错误：无法导入名称'；SynapseCompute'；

嗨，我在运行导入时遇到此错误

从azureml.pipeline.steps导入ParallelRunConfig，ParallelRunStep

错误是：

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-10-a2ef999ba35b> in <module>
----> 1 from azureml.pipeline.steps import ParallelRunConfig, ParallelRunStep
      2 from azureml.pipeline.core import PipelineData
      3 
      4 default_ds = ws.get_default_datastore()
      5 

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/azureml/pipeline/steps/__init__.py in <module>
     38 from .parallel_run_step import ParallelRunStep
     39 from .kusto_step import KustoStep
---> 40 from .synapse_spark_step import SynapseSparkStep
     41 
     42 __all__ = ["AdlaStep",

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/azureml/pipeline/steps/synapse_spark_step.py in <module>
      5 """Contains functionality to create an Azure ML Synapse step that runs Python script."""
      6 from azureml._base_sdk_common._docstring_wrapper import experimental
----> 7 from azureml.pipeline.core._synapse_spark_step_base import _SynapseSparkStepBase
      8 
      9 

/anaconda/envs/azureml_py36/lib/python3.6/site-packages/azureml/pipeline/core/_synapse_spark_step_base.py in <module>
      5 """Contains functionality to create an Azure ML Synapse step that runs Python script."""
      6 
----> 7 from azureml.core.compute import SynapseCompute
      8 from azureml.core.runconfig import RunConfiguration
      9 from azureml.data.constants import DIRECT_MODE

ImportError: cannot import name 'SynapseCompute'

---------------------------------------------------------------------------
ImportError回溯（最近一次呼叫最后一次）
在里面
---->1从azureml.pipeline.steps导入ParallelRunConfig，ParallelRunStep
2从azureml.pipeline.core导入PipelineData
3.
4 default\u ds=ws.get\u default\u datastore（）
5.
/anaconda/envs/azureml_py36/lib/python3.6/site packages/azureml/pipeline/steps/__init__;.py in
38从.parallel\u run\u步骤导入ParallelRunStep
39 from.kusto_步骤导入KustoStep
--->40来自synapse\u spark\u步骤导入SynapseParkStep
41
42 uuu all uuuuu=[“AdlaStep”，
/anaconda/envs/azureml_py36/lib/python3.6/site-packages/azureml/pipeline/steps/synapse_spark_step.py in
5“包含创建运行Python脚本的Azure ML Synapse步骤的功能。”“”
6来自azureml.\u base\u sdk\u common.\u docstring\u包装导入
---->7来自azureml.pipeline.core.\u synapse\u spark\u step\u base导入\u SynapseParkStepBase
8.
9
/anaconda/envs/azureml\u py36/lib/python3.6/site-packages/azureml/pipeline/core//u synapse\u spark\u step\u base.py in
5“包含创建运行Python脚本的Azure ML Synapse步骤的功能。”“”
6.
---->7从azureml.core.compute导入SynapseCompute
8从azureml.core.runconfig导入RunConfiguration
9从azureml.data.constants导入直接_模式
ImportError:无法导入名称“SynapseCompute”

您是否在笔记本中更新了AzureML？我运行了以下

！pip安装--升级AzureML sdk

，该程序最初由于依赖项冲突而失败。之后，我收到了与您相同的错误。重新启动计算，再次升级AzureML sdk（尽管未安装任何软件包）在那之后错误就消失了。可能是在升级时某个正在运行的内核的内存中堆积了什么东西