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.
import onnx fails due to version difference with caffe2 when installing using conda
See original GitHub issueWhen I use Anaconda3 to install both Caffe2 and Onnx ‘import onnx’ fails. However, onnx by itself works fine, its only when paired with Caffe2 that there is a problem.
Below I have included steps to reproduce the error. The first test demonstrates that onnx alone is fine. The second test pairs it with caffe2 to cause the import error.
I would appreciate any insights or ideas to try to resolve the problem.
Thanks, Michael
System
Ubuntu 16.04LTS
Anaconda3 version
conda list anaconda
# packages in environment at /home/mmajursk/anaconda3:
#
# Name Version Build Channel
anaconda custom py36hbbc8b67_0
anaconda-client 1.6.9 py36_0
anaconda-navigator 1.7.0 py36_0
anaconda-project 0.8.2 py36h44fb852_0
Conda version conda --version conda 4.5.4
Sanity Test of just onnx
# Setup new virtual environment for testing
conda create --name test
conda activate test
# install onnx
conda install -c conda-forge onnx
# check install
conda list
# packages in environment at /home/mmajursk/anaconda3/envs/test:
# Name Version Build Channel
ca-certificates 2018.4.16 0 conda-forge
certifi 2018.4.16 py35_0 conda-forge
libgcc-ng 7.2.0 hdf63c60_3
libgfortran-ng 7.2.0 hdf63c60_3
libopenblas 0.2.20 h9ac9557_4
libprotobuf 3.5.2 0 conda-forge
libstdcxx-ng 7.2.0 hdf63c60_3
ncurses 5.9 10 conda-forge
numpy 1.14.3 py35h28100ab_1
numpy-base 1.14.3 py35h0ea5e3f_1
onnx 1.1.2 py35h0c63530_0 conda-forge
openssl 1.0.2o 0 conda-forge
pip 9.0.3 py35_0 conda-forge
protobuf 3.5.2 py35_0 conda-forge
python 3.5.5 1 conda-forge
readline 7.0 0 conda-forge
setuptools 39.2.0 py35_0 conda-forge
six 1.11.0 py35_1 conda-forge
sqlite 3.20.1 2 conda-forge
tk 8.6.7 0 conda-forge
wheel 0.31.0 py35_0 conda-forge
xz 5.2.3 0 conda-forge
zlib 1.2.11 0 conda-forge
Test onnx import
python -c 'import onnx'
Success 😃
Test with caffe2
conda create --name testcaffe2
conda activate testcaffe2
# install caffe2 with the correct version of cuda and cudnn
conda install -c caffe2 caffe2-cuda8.0-cudnn7
# check install
conda list caffe2
# Name Version Build Channel
caffe2-cuda8.0-cudnn7 0.8.dev py36_2018.05.14 caffe2
# install onnx via anaconda
conda install -c conda-forge onnx
# check install
conda list onnx
# Name Version Build Channel
onnx 1.0.1 py36_1 conda-forge
Test onnx import
python -c 'import onnx'
# Results in
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/home/mmajursk/anaconda3/envs/testcaffe2/lib/python3.6/site-packages/onnx/__init__.py", line 10, in <module>
import onnx.helper # noqa
File "/home/mmajursk/anaconda3/envs/testcaffe2/lib/python3.6/site-packages/onnx/helper.py", line 15, in <module>
import onnx.defs as defs
File "/home/mmajursk/anaconda3/envs/testcaffe2/lib/python3.6/site-packages/onnx/defs/__init__.py", line 6, in <module>
import onnx.onnx_cpp2py_export.defs as C
ImportError: /home/mmajursk/anaconda3/envs/testcaffe2/lib/python3.6/site-packages/onnx/onnx_cpp2py_export.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZNK6google8protobuf7Message11GetTypeNameEv
Check complete conda package list
conda list
# packages in environment at /home/mmajursk/anaconda3/envs/testcaffe2:
#
# Name Version Build Channel
blas 1.0 mkl
bzip2 1.0.6 h14c3975_5
ca-certificates 2018.4.16 0 conda-forge
caffe2-cuda8.0-cudnn7 0.8.dev py36_2018.05.14 caffe2
cairo 1.14.12 h7636065_2
certifi 2018.4.16 py36_0 conda-forge
ffmpeg 3.4 h7264315_0
fontconfig 2.12.6 h49f89f6_0
freeglut 3.0.0 hf484d3e_5
freetype 2.8 hab7d2ae_1
future 0.16.0 py36_1
gflags 2.2.1 hf484d3e_0
glib 2.56.1 h000015b_0
glog 0.3.5 hf484d3e_1
graphite2 1.3.11 h16798f4_2
harfbuzz 1.7.6 h5f0a787_1
hdf5 1.10.1 h9caa474_1
icu 58.2 h9c2bf20_1
intel-openmp 2018.0.0 8
jasper 2.0.14 h07fcdf6_1
jpeg 9b h024ee3a_2
libedit 3.1.20170329 h6b74fdf_2
libffi 3.2.1 hd88cf55_4
libgcc-ng 7.2.0 hdf63c60_3
libgfortran-ng 7.2.0 hdf63c60_3
libglu 9.0.0 h0c0bdc1_1
libopus 1.2.1 hb9ed12e_0
libpng 1.6.34 hb9fc6fc_0
libprotobuf 3.5.2 h6f1eeef_0
libstdcxx-ng 7.2.0 hdf63c60_3
libtiff 4.0.9 he85c1e1_1
libvpx 1.6.1 h888fd40_0
libxcb 1.13 h1bed415_1
libxml2 2.9.8 h26e45fe_1
mkl 2018.0.2 1
mkl_fft 1.0.1 py36h3010b51_0
mkl_random 1.0.1 py36h629b387_0
ncurses 6.1 hf484d3e_0
numpy 1.14.3 py36hcd700cb_1
numpy-base 1.14.3 py36h9be14a7_1
onnx 1.0.1 py36_1 conda-forge
opencv 3.3.1 py36h61133dd_2
openssl 1.0.2o 0 conda-forge
pcre 8.42 h439df22_0
pip 10.0.1 py36_0
pixman 0.34.0 hceecf20_3
protobuf 3.5.2 py36hf484d3e_0
python 3.6.5 hc3d631a_2
readline 7.0 ha6073c6_4
setuptools 39.1.0 py36_0
six 1.11.0 py36h372c433_1
sqlite 3.23.1 he433501_0
tk 8.6.7 hc745277_3
wheel 0.31.1 py36_0
xz 5.2.4 h14c3975_4
zlib 1.2.11 ha838bed_2
onnx version differences between conda install results
Based on a diff on the two ‘conda list’ results there is a conflict with:
libprotobuf
(test)libprotobuf 3.5.2 0 conda-forge
(testcaffe2)libprotobuf 3.5.2 h6f1eeef_0
protobuf
(test)protobuf 3.5.2 py35_0 conda-forge
(testcaffe2)protobuf 3.5.2 py36hf484d3e_0
onnx
(test)onnx 1.1.2 py35h0c63530_0 conda-forge
(testcaffe2)onnx 1.0.1 py36_1 conda-forge
If I try to force conda to install onnx v1.1.2 within testcaffe2 environment conda tries to downgrade everything to python2.7
conda install -c conda-forge onnx=1.1.2
Solving environment: done
## Package Plan ##
environment location: /home/mmajursk/anaconda3/envs/testcaffe2
added / updated specs:
- onnx=1.1.2
The following packages will be downloaded:
package | build
---------------------------|-----------------
certifi-2018.4.16 | py27_0 142 KB conda-forge
protobuf-3.5.2 | py27_0 1.3 MB conda-forge
numpy-base-1.14.3 | py27h9be14a7_1 4.1 MB
six-1.11.0 | py27_1 20 KB conda-forge
wheel-0.31.0 | py27_0 61 KB conda-forge
opencv-3.3.1 | py27h61133dd_2 38.9 MB
future-0.16.0 | py27_0 665 KB conda-forge
mkl_random-1.0.1 | py27_0 1.1 MB conda-forge
caffe2-cuda8.0-cudnn7-0.8.dev| py27_2018.05.14 30.0 MB caffe2
numpy-1.14.3 | py27hcd700cb_1 41 KB
mkl_fft-1.0.2 | py27_0 466 KB conda-forge
setuptools-39.2.0 | py27_0 582 KB conda-forge
pip-9.0.3 | py27_0 1.8 MB conda-forge
python-2.7.15 | h1571d57_0 12.1 MB
onnx-1.1.2 | py27h0c63530_0 7.9 MB conda-forge
------------------------------------------------------------
Total: 99.2 MB
The following NEW packages will be INSTALLED:
onnx: 1.1.2-py27h0c63530_0 conda-forge
The following packages will be UPDATED:
ca-certificates: 2018.03.07-0 --> 2018.4.16-0 conda-forge
caffe2-cuda8.0-cudnn7: 0.8.dev-py36_2018.05.14 caffe2 --> 0.8.dev-py27_2018.05.14 caffe2
certifi: 2018.4.16-py36_0 --> 2018.4.16-py27_0 conda-forge
mkl_fft: 1.0.1-py36h3010b51_0 --> 1.0.2-py27_0 conda-forge
mkl_random: 1.0.1-py36h629b387_0 --> 1.0.1-py27_0 conda-forge
numpy: 1.14.3-py36hcd700cb_1 --> 1.14.3-py27hcd700cb_1
numpy-base: 1.14.3-py36h9be14a7_1 --> 1.14.3-py27h9be14a7_1
opencv: 3.3.1-py36h61133dd_2 --> 3.3.1-py27h61133dd_2
openssl: 1.0.2o-h20670df_0 --> 1.0.2o-0 conda-forge
protobuf: 3.5.2-py36hf484d3e_0 --> 3.5.2-py27_0 conda-forge
setuptools: 39.1.0-py36_0 --> 39.2.0-py27_0 conda-forge
six: 1.11.0-py36h372c433_1 --> 1.11.0-py27_1 conda-forge
The following packages will be DOWNGRADED:
future: 0.16.0-py36_1 --> 0.16.0-py27_0 conda-forge
pip: 10.0.1-py36_0 --> 9.0.3-py27_0 conda-forge
python: 3.6.5-hc3d631a_2 --> 2.7.15-h1571d57_0
wheel: 0.31.1-py36_0 --> 0.31.0-py27_0 conda-forge
Proceed ([y]/n)?
Issue Analytics
- State:
- Created 5 years ago
- Comments:8 (3 by maintainers)
Top Results From Across the Web
Install | Caffe2
NOTE: This will install Caffe2 and all of its required dependencies into the current conda environment. We strongly suggest that you create a...
Read more >libcaffeconverter import error for caffe to onnx - Stack Overflow
Workaround: Create a python 3.6 environment with your package manager of choice in the terminal: conda create -n <your env name here> python=3.6 ......
Read more >Onnx Caffe2 - :: Anaconda.org
This library provides Caffe2 importer and exporter for the ONNX format.
Read more >Comprehensive Guide: Installing Caffe2 with GPU Support by ...
If you have not installed CUDA toolkit, are unsure whether it is already installed, or want to install the newer version of CUDA...
Read more >MMCV Installation — MMDetection 2.11.0 documentation
PyTorch/CUDA Environment¶ · Check you GCC version and use GCC 5.4. · Check whether PyTorch is correctly installed and could use CUDA op,...
Read more >
Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free
Top Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
I installed caffe2 using conda. See the top of my first post:
conda create --name testcaffe2 conda activate testcaffe2
install caffe2 with the correct version of cuda and cudnn
conda install -c caffe2 caffe2-cuda8.0-cudnn7
check install
conda list caffe2
Name Version Build Channel
caffe2-cuda8.0-cudnn7 0.8.dev py36_2018.05.14 caffe2
install onnx via anaconda
conda install -c conda-forge onnx
check install
conda list onnx
Name Version Build Channel
onnx 1.0.1 py36_1 conda-forge
Is this still relevant? Closing this now.