无法在Azure服务器上导入AutoMLConfig [英] Cannot import AutoMLConfig on Azure server

问题描述

当我从Azure中运行

When I run

from azureml.train.automl import AutoMLConfig

在Azure中笔记本，我收到以下错误 

in an Azure notebook, I get the following error 

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-14-9b5a7ce91deb> in <module>
----> 1 from azureml.train.automl import AutoMLConfig

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/__init__.py in <module>
     28     # Suppress the warnings at the import phase.
     29     warnings.simplefilter("ignore")
---> 30     from .automl import fit_pipeline
     31     from .automlconfig import AutoMLConfig
     32     from .automl_step import AutoMLStep

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/automl.py in <module>
      8 import pandas as pd
      9 import scipy
---> 10 from automl.client.core.common import data_transformation, fit_pipeline as fit_pipeline_helper
     11 from automl.client.core.common.automl_pipeline import AutoMLPipeline
     12 from automl.client.core.common.datasets import ClientDatasets

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/_vendor/automl/client/core/common/fit_pipeline.py in <module>
     11 import sklearn.pipeline
     12 
---> 13 from . import constants, data_transformation, logging_utilities, pipeline_run_helper, training_utilities, utilities, \
     14     memory_utilities
     15 from .automl_base_settings import AutoMLBaseSettings

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/_vendor/automl/client/core/common/pipeline_run_helper.py in <module>
     15 from automl.client.core.common import logging_utilities as logging_utilities
     16 from automl.client.core.common import metrics
---> 17 from automl.client.core.common import pipeline_spec as pipeline_spec_module
     18 from automl.client.core.common import utilities
     19 from automl.client.core.common.activity_logger import ActivityLogger

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/_vendor/automl/client/core/common/pipeline_spec.py in <module>
     14 from sklearn import linear_model, pipeline
     15 
---> 16 from automl.client.core.common import constants, model_wrappers, tf_wrappers
     17 
     18 PREPROC_NAME = 'preproc'

~/anaconda3_501/lib/python3.6/site-packages/azureml/train/automl/_vendor/automl/client/core/common/tf_wrappers.py in <module>
     47         'crelu': tf.nn.crelu,
     48         'elu': tf.nn.elu,
---> 49         'selu': tf.nn.selu,
     50         'sigmoid': tf.sigmoid,
     51         'softplus': tf.nn.softplus,

AttributeError: module 'tensorflow.python.ops.nn' has no attribute 'selu'

我不知道为什么会出现这个错误以及如何修复它。任何建议将不胜感激。

I have no idea why this error occurs and how to fix it. Any advice will be much appreciated.

推荐答案

您好， 
$


我明白了该错误也出现在我的AutoML Azure笔记本上。 在以下行中：


来自azureml.train.automl的
导入AutoMLConfig



我做了一个pip安装of tensorflow并重新启动内核，错误就消失了。



！pip install tensorflow == 1.13.1



我希望以上帮助。



谢谢 

Hi, 

I see the error appear on my AutoML Azure notebooks as well.  On the following line:

from azureml.train.automl import AutoMLConfig

I did a pip install of tensorflow and restarted the kernel, the error went away.

!pip install tensorflow==1.13.1

I hope the above helps.

Thanks 

这篇关于无法在Azure服务器上导入AutoMLConfig的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！