undefined symbol: _PyThreadState_Current when importing tensorflow

Question

I am on raspberry pi 3. Running Raspbian 9. Python version 3.7.2 installed following https://neoctobers.readthedocs.io/en/latest/rpi/install_python3.html. Made python3 point to python3.7.2 using update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.7 1

This is the error I am getting:

pi@raspberrypi:~ $ python3
Python 3.7.2 (default, May  5 2019, 18:41:29) 
[GCC 6.3.0 20170516] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
Traceback (most recent call last):
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "/usr/local/opt/python-3.7.2/lib/python3.7/imp.py", line 242, in load_module
    return load_dynamic(name, filename, file)
  File "/usr/local/opt/python-3.7.2/lib/python3.7/imp.py", line 342, in load_dynamic
    return _load(spec)
ImportError: /usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so: undefined symbol: _PyThreadState_Current

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/__init__.py", line 24, in <module>
    from tensorflow.python import pywrap_tensorflow  # pylint: disable=unused-import
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/__init__.py", line 49, in <module>
    from tensorflow.python import pywrap_tensorflow
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 74, in <module>
    raise ImportError(msg)
ImportError: Traceback (most recent call last):
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "/usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "/usr/local/opt/python-3.7.2/lib/python3.7/imp.py", line 242, in load_module
    return load_dynamic(name, filename, file)
  File "/usr/local/opt/python-3.7.2/lib/python3.7/imp.py", line 342, in load_dynamic
    return _load(spec)
ImportError: /usr/local/opt/python-3.7.2/lib/python3.7/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so: undefined symbol: _PyThreadState_Current

Thanks in advance.

I have the same problem with python3.7.3.

jaco
– jaco

2019-05-17 18:42:42 +00:00
Commented May 17, 2019 at 18:42 — jaco
– jaco, Commented May 17, 2019 at 18:42

Oleksii · Accepted Answer · 2019-05-23 15:05:50Z

0

There is an approved issue with Tensorflow wheel and Python 3.7 (https://github.com/bennuttall/piwheels/issues/146). Looks like there is no way to solve it gently.

I was managed to successfully install tensorflow 1.13.0 on my RPi3 with Python 3.5.3:

$ python3 --version
Python 3.5.3

$ pip3 freeze | grep tensorflow
tensorflow==1.13.1
tensorflow-estimator==1.13.0

This setup works, but tensorflow raises warnings related to runtime version compatibility:

$ python3 -c "import tensorflow as tf; tf.enable_eager_execution(); print(tf.reduce_sum(tf.random_normal([1000, 1000])))"
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: compiletime version 3.4 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.5
  return f(*args, **kwds)
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: builtins.type size changed, may indicate binary incompatibility. Expected 432, got 412
  return f(*args, **kwds)
tf.Tensor(624.73706, shape=(), dtype=float32)

You may try to live with those warnings (suppress them) or consider alternative tensorflow build for RPi from this repo https://github.com/lhelontra/tensorflow-on-arm/releases (looks like it's available for Python 3.5 and 2.7 only):

$ pip3 install https://github.com/lhelontra/tensorflow-on-arm/releases/download/v1.13.1/tensorflow-1.13.1-cp35-none-linux_armv7l.whl

In my case it works fine without any warnings:

$ python3 -c "import tensorflow as tf; tf.enable_eager_execution(); print(tf.reduce_sum(tf.random_normal([1000, 1000])))"
tf.Tensor(448.25854, shape=(), dtype=float32)

edited May 23, 2019 at 15:05

answered May 23, 2019 at 14:54

Oleksii

3042 gold badges7 silver badges16 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

binaryDragon Over a year ago

Yes I faced these warnings and had decided to live with it. But then tflearn had problems with it. That is why I decided to try Python3.7. But a while ago, on a magical 10th ish reinstall it worked. No idea why. I did what you answered too! Had come to delete this question when I saw this answer. Thanks for the help. Marking it as correct.

Collectives™ on Stack Overflow

undefined symbol: _PyThreadState_Current when importing tensorflow

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related