keras: Import Error with Tensorflow as Backend

I ran with the latest version of tensorflow and keras. I didn’t experience an import problem with theano as the backend. And when I put tensorflow as the backend, I saw the following message:

` In [1]: import keras Using TensorFlow backend.

AttributeError Traceback (most recent call last) <ipython-input-1-c74e2bd4ca71> in <module>() ----> 1 import keras

//anaconda/lib/python2.7/site-packages/keras/init.py in <module>() 1 from future import absolute_import 2 ----> 3 from . import activations 4 from . import applications 5 from . import backend

//anaconda/lib/python2.7/site-packages/keras/activations.py in <module>() 4 from . import backend as K 5 from .utils.generic_utils import deserialize_keras_object ----> 6 from .engine import Layer 7 8

//anaconda/lib/python2.7/site-packages/keras/engine/init.py in <module>() 6 from .topology import Layer 7 from .topology import get_source_inputs ----> 8 from .training import Model

//anaconda/lib/python2.7/site-packages/keras/engine/training.py in <module>() 22 from … import metrics as metrics_module 23 from …utils.generic_utils import Progbar —> 24 from … import callbacks as cbks 25 from …legacy import interfaces 26

//anaconda/lib/python2.7/site-packages/keras/callbacks.py in <module>() 24 if K.backend() == ‘tensorflow’: 25 import tensorflow as tf —> 26 from tensorflow.contrib.tensorboard.plugins import projector 27 28

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/init.py in <module>() 28 from tensorflow.contrib import deprecated 29 from tensorflow.contrib import distributions —> 30 from tensorflow.contrib import factorization 31 from tensorflow.contrib import framework 32 from tensorflow.contrib import graph_editor

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/factorization/init.py in <module>() 22 from tensorflow.contrib.factorization.python.ops.clustering_ops import * 23 from tensorflow.contrib.factorization.python.ops.factorization_ops import * —> 24 from tensorflow.contrib.factorization.python.ops.gmm import * 25 from tensorflow.contrib.factorization.python.ops.gmm_ops import * 26 # pylint: enable=wildcard-import

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/factorization/python/ops/gmm.py in <module>() 25 from tensorflow.contrib.framework.python.framework import checkpoint_utils 26 from tensorflow.contrib.framework.python.ops import variables —> 27 from tensorflow.contrib.learn.python.learn.estimators import estimator 28 from tensorflow.contrib.learn.python.learn.estimators import model_fn as model_fn_lib 29 from tensorflow.python.framework import constant_op

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/init.py in <module>() 85 86 # pylint: disable=wildcard-import —> 87 from tensorflow.contrib.learn.python.learn import * 88 # pylint: enable=wildcard-import 89

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/init.py in <module>() 21 22 # pylint: disable=wildcard-import —> 23 from tensorflow.contrib.learn.python.learn import * 24 # pylint: enable=wildcard-import

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/init.py in <module>() 23 from tensorflow.contrib.learn.python.learn import basic_session_run_hooks 24 from tensorflow.contrib.learn.python.learn import datasets —> 25 from tensorflow.contrib.learn.python.learn import estimators 26 from tensorflow.contrib.learn.python.learn import graph_actions 27 from tensorflow.contrib.learn.python.learn import learn_io as io

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/init.py in <module>() 295 from tensorflow.contrib.learn.python.learn.estimators._sklearn import NotFittedError 296 from tensorflow.contrib.learn.python.learn.estimators.constants import ProblemType –> 297 from tensorflow.contrib.learn.python.learn.estimators.dnn import DNNClassifier 298 from tensorflow.contrib.learn.python.learn.estimators.dnn import DNNEstimator 299 from tensorflow.contrib.learn.python.learn.estimators.dnn import DNNRegressor

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/dnn.py in <module>() 27 from tensorflow.contrib.layers.python.layers import optimizers 28 from tensorflow.contrib.learn.python.learn import metric_spec —> 29 from tensorflow.contrib.learn.python.learn.estimators import dnn_linear_combined 30 from tensorflow.contrib.learn.python.learn.estimators import estimator 31 from tensorflow.contrib.learn.python.learn.estimators import head as head_lib

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/dnn_linear_combined.py in <module>() 29 from tensorflow.contrib.layers.python.layers import optimizers 30 from tensorflow.contrib.learn.python.learn import metric_spec —> 31 from tensorflow.contrib.learn.python.learn.estimators import estimator 32 from tensorflow.contrib.learn.python.learn.estimators import head as head_lib 33 from tensorflow.contrib.learn.python.learn.estimators import model_fn

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/estimator.py in <module>() 47 from tensorflow.contrib.learn.python.learn.estimators import tensor_signature 48 from tensorflow.contrib.learn.python.learn.estimators._sklearn import NotFittedError —> 49 from tensorflow.contrib.learn.python.learn.learn_io import data_feeder 50 from tensorflow.contrib.learn.python.learn.utils import export 51 from tensorflow.contrib.learn.python.learn.utils import saved_model_export_utils

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/learn_io/init.py in <module>() 19 from future import print_function 20 —> 21 from tensorflow.contrib.learn.python.learn.learn_io.dask_io import extract_dask_data 22 from tensorflow.contrib.learn.python.learn.learn_io.dask_io import extract_dask_labels 23 from tensorflow.contrib.learn.python.learn.learn_io.dask_io import HAS_DASK

//anaconda/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/learn_io/dask_io.py in <module>() 24 try: 25 # pylint: disable=g-import-not-at-top —> 26 import dask.dataframe as dd 27 allowed_classes = (dd.Series, dd.DataFrame) 28 HAS_DASK = True

//anaconda/lib/python2.7/site-packages/dask/dataframe/init.py in <module>() 1 from future import print_function, division, absolute_import 2 ----> 3 from .core import (DataFrame, Series, Index, _Frame, map_partitions, 4 repartition) 5 from .io import (from_array, from_bcolz, from_array, from_bcolz,

//anaconda/lib/python2.7/site-packages/dask/dataframe/core.py in <module>() 36 return_scalar = ‘return_scalar’ 37 —> 38 pd.computation.expressions.set_use_numexpr(False) 39 40

AttributeError: ‘module’ object has no attribute ‘computation’ `

However, I did not experience any error when simply import tensorflow directly:

import tensorflow

About this issue

Original URL
State: closed
Created 7 years ago
Comments: 18 (3 by maintainers)

Most upvoted comments

Just update dask to 0.15.0, the problem will disappear

+13

xfsm1912 on Jun 20, 2017

Thanks so much once I updated dask and pandas, works now

chimera9900 on Sep 26, 2017

My pandas package is updated.

For dask package, before updating it, running from tensorflow.contrib.tensorboard.plugins import projector gives the same error message. But after updating dask, the error message dissappeared, by running this import and import keras (with Tensorflow backend)

It looks like it is resolved.

stephenhky on May 9, 2017

i have keras version== 2.2.4 , tensorflow version == 2.0.0b1, numpy version version == 1.16.1 and i m still getting error

Anjali0208 on Jul 11, 2019

Big problem that wasted a week of mine small solution if keras is giving import problem and you are facing “no module named keras” even if you have installed it.

just upgrade your pip by: python -m pip install –upgrade pip

re install upgraded keras and tensorflow by: 2. pip install keras

pip install tensorflow

Hope it will solve the problem. If not, try upgrading your conda (Anaconda) and then do steps 1 to 3 above again

Qaazi on Apr 28, 2019

Hi I had similar issue and was able to fix by reinstalling TensorFlow using sudo conda install tensorflow

This allows and gave permission to downgrade to the compatible numba: 0.35.0-np113py36_6 --> 0.35.0-np112py36_0 numpy: 1.13.3-py36h2cdce51_0 --> 1.12.1-py36h8871d66_1

rsalomon on Feb 12, 2018

Ran through same issue and applied same fix (update dask). Most likely this was caused by updating pandas to recent version (saw it after updating pandas to 0.20.1).

abedkhooli on May 10, 2017