tensorflow: ImportError: No module named google.protobuf

Problem description This issue happens randomly. For same machine, same CPU,works randomly. I run it for 10 times, 6 times succeed, 4 times fail. Same code is running for every time. The code is

import tensorflow as tf
# with tf.device('/cpu:0'):
a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
c = tf.matmul(a, b)
with tf.Session(config=tf.ConfigProto(log_device_placement=True)) as sess:
	print sess.run(c)

System information Information of cluster: Basically our cluster is made of IBM Power System S822LC for commercial computing

OS: Red Hat Enterprise Linux Server release 7.2 (Maipo)
TF version: 0.10.0rc0
protobuf: 3.0.0b2
Python: 2.7.5

Error

>>> import os
>>> cwd = os.getcwd()
>>> cwd
'/gpfs/gpfs0/groups/duraisamy/shawnpan/test/script'
>>> import tensorflow
import tensorflow
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcublas.so.8.0 locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcudnn.so.5.1.5 locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcufft.so.8.0 locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcuda.so.1 locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcurand.so.8.0 locally
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/gpfs/gpfs0/software/rhel72/packages/tensorflow_gpu/tensorflow/__init__.py", line 23, in <module>
    from tensorflow.python import *
  File "/gpfs/gpfs0/software/rhel72/packages/tensorflow_gpu/tensorflow/python/__init__.py", line 58, in <module>
    raise ImportError(msg)
ImportError: Traceback (most recent call last):
  File "/gpfs/gpfs0/software/rhel72/packages/tensorflow_gpu/tensorflow/python/__init__.py", line 52, in <module>
    from tensorflow.core.framework.graph_pb2 import *
  File "/gpfs/gpfs0/software/rhel72/packages/tensorflow_gpu/tensorflow/core/framework/graph_pb2.py", line 6, in <module>
    from google.protobuf import descriptor as _descriptor
ImportError: No module named google.protobuf


Error importing tensorflow.  Unless you are using bazel,
you should not try to import tensorflow from its source directory;
please exit the tensorflow source tree, and relaunch your python interpreter
from there.

About this issue

Original URL
State: closed
Created 8 years ago
Comments: 28 (5 by maintainers)

Commits related to this issue

Fix for https://github.com/tensorflow/tensorflow/issues/6341 This might really be a bug in protobuf-python, https://github.com/google/protobuf/issues/1296 — committed to akesandgren/easybuild-easyblocks by akesandgren 6 years ago
Update requirements.txt generated via: ```bash conda install pipreqs pipreqs . --force ``` Then need to install protobuf via conda or via hackery with pip https://github.com/tensorflow/tensorflow/iss... — committed to sparks-baird/AMDNet by sgbaird 2 years ago

Most upvoted comments

I install tensorflow with virtualenv. pip uninstall protobuf pip uninstall google pip install google pip install protobuf That is ok,but I don’t know why

+175

k-lam on Nov 1, 2017

the __init__.py file is missing in site-packages/google directory. creating an empty __init__.py file there should work. thanks. @fangbq

+48

malithakabir on Apr 1, 2018

On Mac OS - Installing tensorflow 1.3 - it will automatically remove other protobuf installs and install protobuf 3.4. However, this does not work and neither does installing or downgrading to any other protobuf version.

However I found a solution. Not sure why this works - but on Mac OS this solved it.

pip install google

+35

bmatichuk on Oct 25, 2017

pip install google doesn’t work for me I just pip install google-cloud and it works

+20

joe32140 on Jan 12, 2018

Finally installed with conda conda install protobuf

+11

alexraju91 on Oct 1, 2019

Finally installed with conda conda install protobuf

It was the solution in my case. Thank you. pip was the problem. I am using conda environment. Use conda install!

gangmuk on Sep 17, 2020

Again: Double-check that there is no __init__.py in your site-packages/google folder or delete it if there is. Easier than the uninstall dance which may or may not work. BTW: You probably do NOT want to install the “google” package: That is the Python bindings to google search engine. Why would you want that in this context? Oh and it trashes your namespace s mentioned in https://github.com/tensorflow/tensorflow/issues/6341#issuecomment-553769366

Flamefire on Feb 22, 2020

I’m following this guide and every step worked fine except of the last one, When i try to run the benchmark i get this error.

Traceback (most recent call last):
  File "scripts/tf_cnn_benchmarks/tf_cnn_benchmarks.py", line 25, in <module>
    import tensorflow as tf
  File "/home/zwazel/.local/lib/python3.6/site-packages/tensorflow/__init__.py", line 24, in <module>
    from tensorflow.python import pywrap_tensorflow  # pylint: disable=unused-import
  File "/home/zwazel/.local/lib/python3.6/site-packages/tensorflow/python/__init__.py", line 52, in <module>
    from tensorflow.core.framework.graph_pb2 import *
  File "/home/zwazel/.local/lib/python3.6/site-packages/tensorflow/core/framework/graph_pb2.py", line 6, in <module>
    from google.protobuf import descriptor as _descriptor
ModuleNotFoundError: No module named 'google.protobuf'

I tried all of the above things and nothing worked. Can someone help me or should i open a new issue for this? I’m running it on a Jetson Nano with Ubuntu 18.04 LTS

zwazel on Jan 13, 2020