google.protobuf.pyext error

[Bleach665]

Windows 10 Enterprise, Python 3.8.7, CUDA 11.1, tensorflow-2.4.0-cp38-cp38-win_amd64.whl. CPU 2x Xeon 5660.

Step to reproduce:

• clean install python 3.8.7
• in command prompt:

  python -m venv env
  env\scripts\activate
  python -m pip install --upgrade pip
  pip install tensorflow-2.4.0-cp38-cp38-win_amd64.whl
  python
  import tensorflow

Received errors:

(env) e:\TF>python
Python 3.8.7 (tags/v3.8.7:6503f05, Dec 21 2020, 17:59:51) [MSC v.1928 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf;
2021-01-20 18:16:10.043814: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "e:\TF\env\lib\site-packages\tensorflow\__init__.py", line 41, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "e:\TF\env\lib\site-packages\tensorflow\python\__init__.py", line 41, in <module>
    from tensorflow.python.eager import context
  File "e:\TF\env\lib\site-packages\tensorflow\python\eager\context.py", line 32, in <module>
    from tensorflow.core.framework import function_pb2
  File "e:\TF\env\lib\site-packages\tensorflow\core\framework\function_pb2.py", line 7, in <module>
    from google.protobuf import descriptor as _descriptor
  File "e:\TF\env\lib\site-packages\google\protobuf\descriptor.py", line 48, in <module>
    from google.protobuf.pyext import _message
ImportError: cannot import name '_message' from 'google.protobuf.pyext' (e:\TF\env\lib\site-packages\google\protobuf\pyext\__init__.py)

I tried to fix this with setx PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION cpp pip install --upgrade --force-reinstall protobuf, but it didn't help.

Also tried tensorflow-2.3.0-cp38-cp38-win_amd64.whl, but the same error occurred

[fo40225]

try to use Anaconda3-2020.11

[Bleach665]

Same error with Anaconda3-2020.11-Windows-x86_64.exe and tensorflow-2.4.0-cp38-cp38-win_amd64.whl.

[fo40225]

Did you install the tf on base env?

[Bleach665]

Did you install the tf on base env?

No, I tried in coda created environment conda create -n env python=3.8 and python -m venv env The same error occurs.

In the near future I will try to reproduce this error on a pure virtual machine.

[fo40225]

If you use a new conda virtual environment, maybe you can try to install protobuf form conda instead of pip.

Or you can try to use tf on base env first to see if it work or not.

[Bleach665]

If you use a new conda virtual environment, maybe you can try to install protobuf form conda instead of pip. Or you can try to use tf on base env first to see if it work or not.

It didn't help.

I couldn't reproduce this error on a clean virtual machine, so I'm closing this issue.

@fo40225 , thanks for your participation and advice.

[Bleach665]

For those who come here through the search: Currently, the error described above occurs if an environment variable 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION' is set to 'cpp'. Earlier, this variable fixed same issue. Now it on the contrary causes it.