Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Breaks after upgrading to TensorFlow 1.2.0

See original GitHub issue

While building docker images, noticed that sonnet breaks with the new version of TF 1.2.0.

root@a28ad161d26d:/# python
Python 3.5.2 |Continuum Analytics, Inc.| (default, Jul  2 2016, 17:53:06)
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import sonnet as snt
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/conda/lib/python3.5/site-packages/sonnet/__init__.py", line 109, in <module>
    from sonnet.python.ops.resampler import resampler
  File "/opt/conda/lib/python3.5/site-packages/sonnet/python/ops/resampler.py", line 33, in <module>
    tf.resource_loader.get_path_to_datafile("_resampler.so"))
  File "/opt/conda/lib/python3.5/site-packages/tensorflow/python/framework/load_library.py", line 64, in load_op_library
    None, None, error_msg, error_code)
tensorflow.python.framework.errors_impl.NotFoundError: /opt/conda/lib/python3.5/site-packages/sonnet/python/ops/_resampler.so: undefined symbol: _ZN10tensorflow15shape_inference16InferenceContext15WithRankAtLeastENS0_11ShapeHandleEiPS2_

TensorFlow was installed via pip, sonnet was compiled and installed from sources. Perhaps, updating the TF submodule with headers should solve this.

Issue Analytics

State:
Created 6 years ago
Comments:7

Top GitHub Comments

1reaction

prijateljcommented, Jun 26, 2017

Thank you, Shuolongbj. Fortuneately, your post revealed one slight mishap, and I remembered another. My only errors were I did not checkout r1.2 in tensorflow (I instead checked out master), and when I did the configure files for bazel in Tensorflow, I realize I called the opt flags as command line arguments, rather than specify them in the configure file of tensorflow. This potentially desynchronized the Tensorflow and Sonnet configurations and probably resulted in the error. I can confirm that sonnet 1.2 installs successfully on mac with Tensorflow 1.2.0.

I added these flags as they are when the tensorflow configure file asks for optimizatoin flags to optimize my build for my mac (this is for my personal machine, these flags may no apply specifically to your machine. Note the syntax though ‘-m[flag]’): -mavx -mavx2 -mfma -msse4.2

1reaction

Shuolongbjcommented, Jun 25, 2017

FYI. cd sonnet/tensorflow git checkout r1.2 ./configure bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg pip uninstall tensorflow pip install /tmp/tensorflow_pkg/tensorflow-1.2.0-cp36-cp36m-macosx_10_7_x86_64.whl cd …/ bazel build --config=opt //:install bazel-bin/install /tmp/sonnet_pkg pip uninstall sonnet pip install /tmp/sonnet_pkg/sonnet-1.2-py3-none-any.whl

Top Results From Across the Web

TensorFlow upgrade to 1.0.0 breaking import #7980 - GitHub

Gunan,. I tried all the upgrades in between 0.10.0, 0.11.0, and 0.12.0 and they all have the same problem. 0.9.0 still works when...

How to pip install old version of library(tensorflow)?

In case you want to install a "normal" release. then you can easily use the following command: pip install --upgrade tensorflow==1.3.0.

Migrate to TensorFlow 2

Upgrade your training, evaluation and model saving code to TF2 ... Programmatically upgrade some parts of your TF1.x code to TF2.

Changelog - pip documentation v22.3.1

Only calculate topological installation order, for packages that are going to be installed/upgraded. This fixes an AssertionError that occurred when determining ...

Release 1.3.1

Release 1.2.0 ... This is a breaking change from the behavior of RNNCells in TensorFlow versions ... ensure it raises no errors, and...