python - 在 Anaconda2 中导入 tensorflow : undefined symbol

Question

今天早上我试图在 Anaconda3(python 版本 3.5 和 Ubuntu 16.04)上安装 tensorflow。我使用 conda install tensorflow 安装了 tensorflow。但是，导入 python/jupyter notebook 时效果不佳。

Python 3.5.6 |Anaconda, Inc.| (default, Aug 26 2018, 21:41:56) 
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/work/.conda/envs/tensorflow/lib/python3.5/site-packages/tensorflow/__init__.py", line 22, in <module>
    from tensorflow.python import pywrap_tensorflow  # pylint: disable=unused-import
  File "/home/work/.conda/envs/tensorflow/lib/python3.5/site-packages/tensorflow/python/__init__.py", line 52, in <module>
    from tensorflow.core.framework.graph_pb2 import *
  File "/home/work/.conda/envs/tensorflow/lib/python3.5/site-packages/tensorflow/core/framework/graph_pb2.py", line 6, in <module>
    from google.protobuf import descriptor as _descriptor
  File "/home/work/.conda/envs/tensorflow/lib/python3.5/site-packages/google/protobuf/descriptor.py", line 47, in <module>
    from google.protobuf.pyext import _message
ImportError: /home/work/.conda/envs/tensorflow/lib/python3.5/site-packages/google/protobuf/pyext/_message.cpython-35m-x86_64-linux-gnu.so: undefined symbol: _ZNK6google8protobuf10TextFormat17FieldValuePrinter9PrintBoolEb
>>> 

有没有人遇到过这种情况？我注意到有类似的情况，但该解决方案可能不适用于 Anaconda3。

Answer 1

使用pip安装tensoflow pypi-package

我有完全相同的错误。我的应用程序需要 tensorflow 版本 1.0.0。 tensorflow的conda-package和pypi-package好像是有区别的。所以在<YOUR_CONDA_ENV>卸载tensorflow的conda-package ironement 和 pip-install pipy-package 再次:

# uninstall current tensorflow conda-package 
$ sudo conda uninstall --name <YOUR_CONDA_ENV> tensorflow
# install the pypi-package of tensorflow
$ sudo <YOUR_CONDA_ENV_PATH>/bin/pip install tensorflow==<REQUIRED_VERSION>

在你的情况下 <YOUR_CONDA_ENV>=tensorflow并且您的 conda 环境位置的路径是 <YOUR_CONDA_ENV_PATH>=/home/work/.conda/envs/tensorflow .

确保使用 pip install您的 conda 环境 ( <YOUR_CONDA_ENV_PATH>/bin/pip ) 而不是一般的 pip ( /bin/pip )。如果您还没有在您的环境中安装 pip，只需:

$ sudo conda install --name <YOUR_CONDA_ENV> pip